AI-Enabled Precision Agriculture for Smallholder Farmers

AI-Enabled Precision Agriculture for Smallholder Farmers | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Systematic Review AI-Enabled Precision Agriculture for Smallholder Farmers Sandip Satpati This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8808017/v1 This work is licensed under a CC BY 4.0 License Status: Under Revision Version 1 posted 11 You are reading this latest preprint version Abstract The application of artificial intelligence (AI) to precision agriculture has become a revolutionary tool for improving the productivity, sustainability, and resiliency of smallholder agricultural systems, especially in developing and rural settings. This paper will discuss the use of AI tools, including machine learning-based crop disease detection, yield prediction models, smart irrigation, and decision-support systems, in the context of smallholder farming based on a systematic review of 50 peer-reviewed articles that were published by reputable journals indexed in Elsevier/ScienceDirect, Taylor and Francis, Wiley, SAGE, and Springer Nature. The review also provides consistent evidence that AI applications have the potential to positively transform agricultural productivity by diagnosing diseases earlier, better using inputs, and managing farms based on data, as well as enhancing environmental sustainability. Nonetheless, the adoption by the smallholder farmers is still disproportionate and highly contextual. There are perceptions and intent to adopt AI-based technologies that are behavioral and socio-economic in nature and are strongly influenced by perceptions and attitudes to AI systems, trust, digital literacy, access, and cost of data infrastructure, and institutional support. The literature also points to the existence of severe obstacles like low levels of connectivity, skills gaps, disjointed extension services, ethical and data-governance issues, and the inaccessibility of high-tech solutions and smallholder realities. Meanwhile, the uptake and impact can be greatly improved with the help of the enabling factors, such as human-centered design, advisory and extension services, facilitating policies, and inclusive innovation ecosystems. This study contributes to the holistic insight into AI-enabled precision agriculture among smallholders by incorporating both technical performance evidence and socio-economic as well as policy viewpoints. It offers a conceptual basis and a research focus for future empirical investigations, stating that future AI solutions to optimize the advantages of digital agriculture must be context-aware, equitable, and farmer-focused to make sure that the advantages of digital agriculture are widely distributed across rural communities. Artificial intelligence in agriculture Precision agriculture Smallholder farmers Technology adoption Sustainable agricultural productivity Figures Figure 1 Figure 2 1. Introduction 1.1 Background and Rationale Smallholder agriculture has continued to be the mainstay of food production in most developing and emerging economies, yet it is increasingly being limited by a complex of structural and environmental issues. The low productivity of smallholder farms has not increased with a lack of access to quality inputs, poor extension services, minor landholdings, and information asymmetry (Aijaz et al., 2025; Mana et al., 2024). The climate change risks such as heightened temperatures, irregular rainfall trends, the increase in pest and disease cases, and soil erosion also compound these limitations and burden the farmers who are resource-constrained (Wu and Zhong, 2025). Moreover, rural labor unavailability caused by urban migration and demographic shifts is exerting a strain on the traditional farm system that is labor-intensive, lowering the efficiency of operations and the ability to manage farms on time (Omotayo et al., 2025). All these problems jeopardize the food security, agricultural incomes, and the sustainability of the agricultural systems based on smallholders, which highlights the necessity of more scalable, adaptive, and data-informed responses. Precision agriculture (PA) has become one of the most significant paradigms currently that utilize data-driven technologies to empower agricultural decision-making on a fine spatial and temporal level. Recent developments in artificial intelligence (AI), machine learning (ML), computer vision, Internet of Things (IoT), and unmanned aerial vehicles (UAVs), have made considerable changes to the functionality of PA systems, allowing them to monitor in real-time, engage in predictive analytics, and perform automated operations in the field (Bayar et al., 2025; Rashid et al., 2025). The AI-powered PA applications now cover such crucial areas as the detection of crop diseases, yield forecasting, intelligent irrigation, nutrient control, and the evaluation of soil health (Majdalawieh et al., 2025; Mamabolo et al., 2025). Combining the heterogeneous sources of data, such as satellite imagery, proximal sensors, weather data, and farm management records, AI systems can create actionable information to enhance the efficiency of resource use, decrease the input costs, and become more resilient to climatic variability (Aijaz et al., 2025; Padhiary et al., 2024). When considering the aspect of sustainability, AI-based PA can help with environmentally friendly farming by reducing the overuse of fertilizers and pesticides and ensuring or even enhancing productivity (Mana et al., 2024). As a result, AI-enabled PA is becoming one of the foundations of climate-smart and sustainable agriculture. Although it has stood to promise great success, the application of precision agriculture has been historically uneven within large-scale commercial operations situated in developed countries where capital accessibility, digital connectivity, and technical know-how are comparatively high (Saiz et al., 2026). Those systems can be based on expensive sensors, custom platforms, sophisticated equipment, and sophisticated data analytics pipelines that can be inaccessible by the majority of smallholder farmers. Subsequently, technological disparity remains, which restricts the inclusiveness and equal influence of PA innovations (Omotayo et al., 2025). As a contrast, smallholder-oriented AI systems of the future are based on affordability, scalability, and relevance to situations. These systems are moving towards more extensive utilization of low-cost IoT sensors, mobile-based decision support systems, small-scale machine learning models, and cloud-based analytics based on small plots and small-scale farming heterogeneity (Bayar et al., 2025; Wu and Zhong, 2025). In addition, the significance of socio-technical design, usability formulated by farmers, and policy facilitation to guarantee meaningful adoption and long-term effects among smallholder groups has been identified in recent studies (Saiz et al., 2026; Wu and Zhong, 2025). This distinction is important to understand to be able to align AI-enabled precision agriculture with the objectives of inclusive development and make sure that the technology will be converted into the actual benefits that smallholder farmers will receive. 1.2 Problem Statement Although there has been rapid growth and development of artificial intelligence (AI) technologies in agriculture, their uptake among the smallholder farmers is widely uneven and disjointed in terms of regional and production systems. Although AI-based solutions are expected to enhance productivity, sustainability, and resilience, empirical data show that such benefits are over- and under-represented in technologically advanced and well-resourced farming environments (Ahmad et al., 2025; Özoğul, 2025). Still, smallholder farmers are grappling with a lack of digital infrastructure, cost levels of implementation, the lack of technical assistance, and the lack of confidence in algorithmic decision-making, which leads to an ongoing adoption gap between the potential of the technology and its application on the ground (Hiywotu, 2025; Sood et al., 2022). The major weakness of the current literature is that most studies are based on technology-focused research methods that emphasize approximately accuracy of algorithms, system architecture, and computing capabilities with minimal consideration on socio-economic, behavioral and institutional aspects of adoption. Most of the studies measure AI systems based on technical performance scales, and little of them incorporate the views of a farmer, contextual limitations, or governance factors (Baladraf et al., 2025; Ryan et al., 2023). This imbalance limits the practical applicability of AI innovations because the decisions of adopting such innovations in smallholder agriculture are not only influenced by the technological effectiveness of the technology but also social pressure, usefulness, risk-taking behavior, and conditions of implementation (Lee et al., 2024; Pearson, 2025). Furthermore, the lack of holistic and integrative frameworks, which directly connect the performance of AI systems to the perceptions of farmers and the consequences of subsequent adoptions, is significant. The available research usually investigates these aspects separately, neglecting dynamic relationships amid technological dependability, trust on behalf of the users, and behavioral intention (Ugwu et al., 2025). Participatory and human-centred design methods that have demonstrated the potential to enhance usability and acceptance are not well represented in AI-for-agriculture studies, especially in the smallholder setting (Su et al., 2026). Due to this, the absence of consistent models that can link AI performance, social-economic perception, and adoption behavior constrains theoretical progress as well as policy implications. The way to bridge these gaps is adhering to interdisciplinary and a socio-technical research paradigm that incorporates the performance assessment of AI with behavioral, economic, and institutional assessments. In the absence of such integrative structures, AI based agricultural solutions can strengthen existing inequalities instead of becoming part of inclusive and sustainable food systems (Ahmad et al., 2025; Ryan et al., 2023). Therefore, the systematic association between AI technological opportunities and farmer-centered adoption processes is urgently required, which would guide scalable, equitable, and context-driven deployment policies in smallholder agriculture. 1.3 Research Objectives I. To assess the impact of AI tools (e.g., crop disease detection, yield prediction) on agricultural productivity. II. To evaluate how rural farmers perceive and adopt AI-based agricultural technologies. III. To identify barriers and enablers for implementing AI in smallholder agriculture. 1.4 Originality and Innovation of the Study. The current study contributes to the current body of research on artificial intelligence (AI) in agriculture by going beyond the largely technology-focused views and providing a comprehensive, smallholder-focused, and policy-relevant contribution. Although the previous knowledge and practice on the technical potential of AI and machine learning to improve productivity, disease management, and resource efficiency have been well documented (Clay et al., 2024; Gupta et al., 2025; Minhans et al., 2025), the practical applicability of technology efficiency to specific change behavior in the real-world setting of smallholder farmers has been relatively overlooked. This paper fills this gap by explicitly relating the effectiveness of AI tools to human adoption processes, which will further facilitate a more socio-technical interpretation of AI-enabled agriculture. First, the research combines AI performance indicators and human-related adoption elements, including trust, perceived usefulness, usability and contextual limits. The literature usually analyzes the AI systems independently, focusing on accuracy, optimization efficiency, or computational progress (Parganiha & Verma, 2025; Adinarayana et al., 2024). Conversely, this study contributes to the growing arguments around the importance of trust-aware and user-friendly AI implementation to agriculture by systematically relating system-level efficiency with agricultural perception and behavioral outcome (Gardezi et al., 2023; Erickson and Fausti, 2021). In such a way, this piece of research will offer both empirical and theoretical data as to why AI solutions that are technically sound can still perform poorly in terms of uptake in the context of smallholders. Second, the research suggests a Smallholder-Centric AI Adoption Framework, which is one of the theoretical and methodological novelties. In contrast to the generic precision agriculture or smart farming frameworks that are mostly developed based on large-scale and capital-intensive farming systems (Clay et al., 2024; Cardak et al., 2025), the given framework explicitly considers the socio-economic context of the smallholder farmers, such as resource availability, digital literacy, institutional backing, and the extension procedures. Based on the findings of the digital technology adoption research carried out in the context of smallholder, the conceptual framework defines adoption as the dynamic process connecting the performance of AI systems, perception of farmers and adoption outcomes. The contribution is a direct reaction to the calls of interdisciplinary and inclusive AI research in agriculture (Gardezi et al., 2023; Gupta et al., 2025). Third, the research provides policy and design recommendations in practice that go beyond the study of algorithmic innovation. Though a large part of the current body of literature focuses on the technological progress, there are not many studies that convert their results into the recommendations to policymakers, designers, and extension systems (Erickson and Fausti, 2021). This study claims that deficiency by creating design concepts of AI tools suited to smallholder, such as affordability, transparency, and compatibility with extension services, and by explaining policy directions to facilitate inclusive AI diffusion by capacity building, governance systems, and investing in digital infrastructures (Ishore et al., 2025; Tran Cao Uy et al., 2024). In this way, the research helps to fill the gap between the agricultural policy and sustainable development goals and AI research. In general, this study is novel in its combination of a socio-technical approach, a specific study of smallholder farmers, and an emphasis on transforming AI innovation into inclusive and scalable agricultural transformation. By matching AI performance and adoption dynamics with the relevance of policies, the research advances the existing body of work on precision agriculture and AI-for-agriculture toward more balanced, impactful results. 2. Precision Agriculture: The Conceptual Underpinnings of AI-Assisted Agriculture. The innovation of Precision Agriculture and Smallholder Farming Systems is the second one. Precision agriculture (PA): This is the use of digital, data-driven, and sensor-based technologies to optimise agricultural inputs and management decisions on a fine spatial and temporal scale. PA, which started as a part of the large-scale mechanized farming systems, now includes artificial intelligence (AI), Internet of Things (IoT), robotics, and big data analytics to support real-time decisions (Atasoy, 2025 ; Abdulraheem et al., 2026 ). Although such innovations yield potential efficiency gains and benefits that are sustainable, their creation has been influenced in large measure by the requirements of industrial farming systems. The problem of structural differences between the smallholder and industrial farming systems is a fundamental factor affecting the applicability of PA technologies. Industrial farms are usually located on large, uninterrupted areas of land, highly mechanized and capitalized and equipped with digital infrastructures, which allow a smooth connection of sophisticated PA instruments (Almazmomi, 2025 ). Conversely, smallholder systems are typified by fragmented plots, diversified production activities, low levels of mechanization, and use of family labor that makes it hard to direct transfer conventional PA models (Nguyen et al., 2023 ; Lankamo et al., 2025 ). There are also limitations that restrict the PA adoption by the smallholders. Poor access to capital and credit will limit investment in sensors, equipment and connectivity, and poor digital infrastructure and network coverage will also complicate the implementation of technology (Sharma et al., 2025). The lack of skills and digital illiteracy limits access to information and access to and trust in data-driven recommendations, which highlights the role played by extension and capacity-building mechanisms (Daum, 2025 ; High et al., 2025 ). Moreover, digital instruments are not usually created to fit the context of a smallholder and are therefore less usable and relevant (Dittmer et al., 2025 ). These limitations lead to the necessity of adaptive, inclusive, and smallholder-oriented PA frameworks. 2.2 Artificial Intelligence in Precision Agriculture Modern precision agriculture has established artificial intelligence (AI) as a core element to optimize crop management and resource utilization based on data. Both machine learning (ML) and deep learning (DL) are utilized all over the yield prediction, soil property estimation, climate risk modeling, and input optimization by learning complex, non-linear relationships using multi-source data in agriculture (Aijaz et al., 2025 ; Ugwu et al., 2025 ). Recent developments of deep neural networks have also increased prediction capability and scalability, application in adaptive decision-making in changing field conditions (Mana et al., 2024 ; Wu and Zhong, 2025 ). One of the most developed uses of AI in precision agriculture is computer vision, in crop disease and pest detection. Vision-based classifiers and convolutional neural networks allow the timely detection of diseases in the form of images of leaves and field imagery and shorten the time of diagnosis, and decrease the losses of yields (Majdalawieh et al., 2025 ; Pearson, 2025 ). These strategies are being used more and more on mobile gadgets and edge computers and enhance their usability in the real-time field (Padhiary et al., 2024 ; Su et al., 2026 ). The combination of AI and the Internet of Things (AIoT), unmanned aerial vehicles (UAVs), proximal sensors, and new 5G connectivity have greatly expanded the space and time resolution of farm surveillance. AIoT can be used to facilitate smart irrigation, nutrient control, and stress detection using sustained data flows of distributed sensors and aerial stations (Bayar et al., 2025 ; Rashid et al., 2025 ). The affordability of sensors, interoperability, and connectivity are, however, major challenges, especially in the smallholder setting (Saiz et al., 2026 ). These technologies meet at AI-based decision support systems and advisory systems that turn complex analytics into actionable recommendations. These systems are vital in the process of bridging the technology proficiency with farmer choice, focusing on usability, confidence, and socio-technical integration (Lee et al., 2024 ; Ahmad et al., 2025 ; Ryan et al., 2023 ). 2.3 Theoretical Lenses for Technology Adoption Adoption of technologies based on artificial intelligence (AI) into agriculture can be viewed in a number of complementary theoretical perspectives that reflect the individual, social and systemic aspects of technological change. The Technology Acceptance Model (TAM) is an initial system that offers a basis with the application of the perceived usefulness and perceived ease of use as key factors that determine technology adoption. Empirical research in the field of agricultural digitalization has shown that the intention to use precision and AI-based applications is largely affected by the expectation of its productivity, income growth, and the decreased amount of labor and risks by farmers (Nguyen et al., 2023 ; Sharma et al., 2025). Nevertheless, perceived ease of use is still limited by digital skills and access, especially in the case of smallholder farmers (Daum, 2025 ; Dittmer et al., 2025 ). In addition to the perceptions of an individual, Diffusion of Innovations theory places emphasis on social learning, communication medium, and institutionalization in influencing the adoption pathways. The impact of agricultural extension services, peer networks, and demonstration effects is substantial regarding the issue of the spread of innovations in the world of farmers (Lee et al., 2024 ; Lankamo et al., 2025 ). Research on AI-assisted extension systems also suggests that the process of diffusion is hastened when technologies are consistent with local practices and promote transformative learning and do not involve mere transfer of information (High et al., 2025 ). The newer research focuses on human-centered AI and socio-technical systems, acknowledging that the use of technology is within a wider social, cultural, and organizational framework. Human-centered design models are focused on usability, inclusivity, and contextual relevance, which means that AI systems assist farmers in decision-making, but do not take the decision-making powers away (Ryan et al., 2023 ; Su et al., 2026 ). Socio-technical approaches also emphasize the interrelationship between technological systems, governing relationships, and user capacities, especially in agricultural regimes characterised by resource-based constraints (Atasoy, 2025 ; Baladraf et al., 2025 ). Lastly, the concepts of trust, explainability, and ethics have become important factors in AI adoption in agriculture. Issues of data ownership, algorithmic secrecy, environmental sustainability as well as social fairness may erode user confidence unless properly resolved (Ahmad et al., 2025 ; Özoğul, 2025 ). Responsible and inclusive adoption of explainable and transparent AI systems with the help of ethical governance frameworks and secure data architectures is thus important (Almazmomi, 2025 ; Abdulraheem et al., 2026 ; Neethirirajan, 2025). These theoretical perspectives have a synthesis that offers a wholesome framework of analyzing AI uptake in precision agriculture in terms of technical, social, and ethical aspects. 3. Methodology 3.1 Data Sources and Selection Criteria. The review had a systematic and clear literature selection process that guaranteed the inclusion of all scholarly literature on artificial intelligence (AI)-enabled precision agriculture with the particular aim of the smallholder farming system and technology adoption. Five large academic publishers and databases with a reputation of high-quality agricultural and interdisciplinary research were searched: Elsevier/ScienceDirect, Taylor and Francis, Wiley Online Library, SAGE Publications and Springer Nature retrieved peer-reviewed journal articles and book chapters. The rationale behind these platforms lies in the fact that they are highly represented in journals listed in the Scopus database and cover the issues of AI, precision agriculture, and rural development research comprehensively (Clay et al., 2024; Ahmad et al., 2025). The period of the review was 20172026, which reflects the dynamism of machine learning, deep learning, AIoT, and socio-technical adoption research in agriculture, especially after the advent of Agriculture 4.0 and 5.0 paradigms (Sood et al., 2022; Baladraf et al., 2025). (Artificial intelligence, precision agriculture, machine learning, smallholder farmers, technology adoption, decision support systems, and AIoT) were some of the keywords that were used to search the material. Inclusion criteria: The studies needed to fulfill the following inclusion criteria: (i) AI-based technologies (e.g., machine learning, deep learning, computer vision, AIoT) were explicitly applied and reviewed; (ii) were located in the precision or smart agriculture contexts; (iii) dealt with adoption, usability, social-economic effects or smallholder agricultural regimes empirically or conceptually. Articles that did not involve the development of the algorithm on the laboratory level and were not followed up with the applications in agriculture, non-peer-reviewed materials, and the articles that did not involve the adoption or farm-level decision-making were excluded. The last corpus contained systematic reviews, empirical researches, and theoretical frameworks that cover AI implementation, governance, ethics, and adoption processes in different agro-ecological and socio-economic settings (Aijaz et al., 2025; Nguyen et al., 2023; Lankamo et al., 2025; Pacal et al., 2024). This strategy provided analytical rigor and also stayed relevant to smallholder-focused precision agricultural research. Table 1: Classification of Reviewed Articles by Major Themes Theme Articles (Author, Year) AI Application / Focus Dimension Addressed Key Findings 1. AI & Technology in Crop Management Aijaz et al., 2025; Bayar et al., 2025; Mamabolo et al., 2025; Majdalawieh et al., 2025; Padhiary et al., 2024; Rashid et al., 2025; Upadhyay et al., 2025; Sudha & Loret, 2026; Odounfa et al., 2025; Filippi et al., 2025; Mahale et al., 2024; Kuradusenge et al., 2024; Ajith et al., 2025; Gupta & Kumar Pal, 2025; Minhans et al., 2025; Parganiha & Verma, 2025; Clay et al., 2024; Gupta et al., 2025 Precision agriculture, ML/DL, UAVs, IoT, sensors, yield prediction, disease detection Technology / System AI improves crop productivity, accuracy of disease detection, and yield prediction across smallholder and commercial contexts 2. Human / Adoption & Smallholder-Centric Factors Omotayo et al., 2025; Wu & Zhong, 2025; Ahmad et al., 2025; Lee et al., 2024; Sood et al., 2022; Su et al., 2026; Tran Cao Uy et al., 2024; Dittmer et al., 2025; High et al., 2025; Nguyen et al., 2023; Sharma et al., 2025; Lankamo et al., 2025 Adoption determinants, participatory design, extension, smallholder engagement Human / System Human factors such as trust, skills, participatory design, and access to digital services significantly influence AI adoption 3. System, Policy & Governance Mana et al., 2024; Saiz et al., 2026; Abdulraheem et al., 2026; Almazmomi, 2025; Atasoy, 2025; Omotayo et al., 2025; Erickson & Fausti, 2021 Policy frameworks, socio-economic challenges, infrastructure, food security System / Technology AI adoption depends on supportive policy, digital infrastructure, and socio-economic context; governance frameworks are essential 4. Sustainability & Socio-Technical Integration Baladraf et al., 2025; Hiywotu, 2025; Ryan et al., 2023; Ahmad et al., 2025; Wu & Zhong, 2025; Mana et al., 2024 Agriculture 4.0/5.0, sustainable agriculture, socio-technical integration Technology / Human / System Integration of AI with human and system factors promotes sustainable agriculture and enhances adoption in developing contexts Source: By Author 3.2 Review Protocol A PRISMA-based (Preferred Reporting Items of Systematic Reviews and Meta-Analyses) screening protocol was used in this review to guarantee the methodological transparency, reproducibility, and rigor in the selection and synthesis of the relevant literature (Figure 1). A preliminary list of records was obtained in Elsevier/ScienceDirect, Taylor & Francis, Wiley, SAGE and Springer Nature databases. Titles and abstracts were filtered after the elimination of duplicates and used to remove irrelevant material according to the criteria of being relevant to AI-enabled precision agriculture, and especially focusing on material relating to smallholder farming systems and technology adoption. The next step implied full-text screening, which ensured that the obtained corpus of peer-reviewed articles complies with the established inclusion criteria, thus a narrowed down set of publications that can be included in qualitative synthesis (Majdalawieh et al., 2025; Pacal et al., 2024). After the screening, thematic coding and synthesis approach was used. Articles were coded inductively and deductively to reflect on recurring conceptual, technical, and socio-institutional concept(s) of AI deployment in precision agriculture. The categories were repeatedly revised to include technological dimensions (e.g., algorithm performance, sensing accuracy) and human-centered dimensions (e.g., adoption drivers, trust, governance), which aligns with socio-technical views highlighted in recent AI-agriculture research (Wu & Zhong, 2025; Ryan et al., 2023). According to thematic convergence, the reviewed articles were categorized as three analytical groups. The former included technical performance studies, which concerned machine learning, deep learning, computer vision, AIoT, UAVs, and sensor-based yield prediction, disease detection, and resource optimization systems (Aijaz et al., 2025; Rashid et al., 2025; Upadhyay et al., 2025). The second group consisted of adoption and perception studies that investigated farmer attitudes and behavior intentions, institutional support, and socio-economic factors that impact the adoption of AI-based precision agriculture, especially in the case of smallholders (Nguyen et al., 2023; Tran Cao Uy et al., 2024; Lankamo et al., 2025). The third group included the study of policy, ethics and sustainability with a focus on governance systems, ethical risks, explainability, equity and long-term sustainability of AI-driven agricultural systems (Omotayo et al., 2025; Ahmad et al., 2025; Gardezi et al., 2023). This systematic procedure facilitated a combined synthesis of technical performance, human adoption procedures, and policy-related findings that is consistent with fresh demands of comprehensive and conscientious AI frameworks in precision agriculture. 3.3 Analytical Framework The analytical model used to develop this research is implemented in the form of a three-layered model, whereby technological, human, and institutional areas are incorporated to determine the adoption and the effect of AI-enabled precision agriculture among smallholder farmers. This framework enables one to understand in depth how AI technologies affect productivity, farmer behavior, and the enabling environment needed to make the adoption sustainable. I. Artificial Intelligence and Productivity Reports. The initial layer dwells upon AI tools and their direct contribution to agricultural productivity. These are machine learning algorithms, computer vision, Unmanned Aerial Vehicles (UAVs), AIoT systems, and sensor networks that detect crop diseases, predict yields, monitor soil, optimize irrigation, and manage pests (Majdalawieh et al., 2025; Bayar et al., 2025; Rashid et al., 2025). Empirical and review literature proves that these tools are more efficient in terms of resources, have less input waste, and provide more efficient interventions to crop health and productivity (Aijaz et al., 2025; Mamabolo et al., 2025; Padhiary et al., 2024). In addition, a combination of AI and IoT and UAV can also be used to enable real-time monitoring and automated decision-making, which can play a role in resilient and sustainable production systems (Mana et al., 2024; Wu and Zhong, 2025). II. Farmer Perception and Adoption Behavior The second level looks into the perceptions, attitudes and adoption behavior of farmers. The perceived usefulness, ease of use, social influence, and economic incentives are essential factors to successful technology uptake, in addition to the technical performance (Lee et al., 2024; Sood et al., 2022; Tran Cao Uy et al., 2024). The awareness, digital literacy, risk perception, and access to the advisory services are the factors that influence the intention of the smallholder farmers to adopt AI-based tools (Nguyen et al., 2023; Su et al., 2026; High et al., 2025). The knowledge of these behavioral determinants is essential in developing human-centered AI systems that will meet the needs, capacities, and local farming practices of farmers (Ryan et al., 2023; Pearson, 2025). III. Policy, Infrastructure and Institutions Enabling Environment. The third layer underlines the enabling environment that facilitates the use of AI in agriculture. This involves the national and regional policies, the digital infrastructure, the institutional capacity, the extension services, and the ethical, transparent, and sustainable AI deployment frameworks (Ahmad et al., 2025; Omotayo et al., 2025; Gardezi et al., 2023). Research points out that regulatory directions, ICT infrastructure investment, and fair access to the technology are essential to implement AI interventions that are inclusive and scalable to smallholders (Daum, 2025; Dittmer et al., 2025; Atasoy, 2025). Also, social-technical integration will promote the fact that AI applications are not merely technologically efficient but also socially and economically feasible in local farming settings (Gupta and Kumar Pal, 2025; Ugwu et al., 2025). Integrated Perspective A combination of these three layers grants the framework a multifaceted perspective to evaluate AI-enabled precision agriculture. It also connects technical performance with adoption conduct and aligns both of them in a conducive institutional setting. In this way, the leverage areas of enhancing productivity, raising adoption rates among smallholders, and advancing sustainable and ethical AI use in the agricultural sector can be identified. 4. Impact of AI Tools on Agricultural Productivity 4.1 AI for Crop Disease Detection CNNs, Vision and UAV-based Monitoring. The latest achievements in the field of artificial intelligence (AI) have made convolutional neural networks (CNNs) and methods of computer vision the fundamental drivers of automated crop disease detection. ResNet, DenseNet, EfficientNet, and customized lightweight architectures are also deployed on the leaf-level and canopy-level classification of disease based on CNN-based architectures because they are highly effective in feature extraction and generalization (Majdalawieh et al., 2025 ; Upadhyay et al., 2025 ; Pacal et al., 2024). RGB, multi spectrum, and hyperspectral based vision systems have been shown to be highly diagnostic when used together with transfer learning and data augmentation methods under controlled and semi-controlled settings (Aijaz et al., 2025 ; Minhans et al., 2025 ). The Unmanned Aerial Vehicles (UAVs) also expand AI-based disease detection capabilities beyond plot-scale surveillance to the field-scale surveillance allowing high-resolution data collection in real-time over large agricultural plots. AI systems using UAVs in combination with CNNs and Internet of Things (IoT) and Artificial Intelligence of Things (AIoT) platforms facilitate the prevention of diseases before they occur, spatial disease mapping, and the use of targeted intervention plans (Bayar et al., 2025 ; Rashid et al., 2025 ). Such systems are becoming more consistent with the precision agriculture paradigms, and dynamic disease-monitoring is possible, with less reliance on labor and the use of chemicals (Gupta and Kumar Pal, 2025 ; Sudha and Loret, 2026). Performance Measures vs. actual scalability. Although CNN-based disease detection models may claim high performance, including accuracy, precision, recall, F1-score, and area under the curve (AUC), their applicability to large-scale learning is limited by the heterogeneity of data sources and their environmental variability and scalability (Majdalawieh et al., 2025 ; Filippi et al., 2025 ). Most models are trained using curated data with homogeneous lighting, backgrounds and growth of crops, thus restricting their robustness when deployed to a wide field of application (Pacal et al., 2024; Wu and Zhong, 2025 ). Computational issues, energy usage, connection density, and rural access to edge or cloud systems also aggravate the problem of scalability (Saiz et al., 2026 ; Omotayo et al., 2025 ). Despite the promising results of lightweight CNNs and edge-AI solutions, trade-offs between the complexity of models and diagnostic accuracy still play a major role (Padhiary et al., 2024 ; Parganiha and Verma, 2025 ). Therefore, recent articles state that explainable, energy-efficient, and context-aware AI models with a focus on balancing the predictive performance and operational feasibility in actual farming settings are needed (Gardezi et al., 2023 ; Özoğul, 2025 ). Asian and African Case Studies. Asian and African experience shows the promise and the constraints of AI-based systems of crop disease detection in smallholder-managed agricultural systems. CNN-based vision systems have been put into use to date with crops in South and Southeast Asia that include rice, tomato, wheat, and chili with significant improvements in the accuracy of identifying diseases and supporting farmers in decision-making (Ajith et al., 2025 ; Su et al., 2026 ; Mahale et al., 2024 ). Nevertheless, digital literacy, the availability of extension services, and socio-economic preparedness are the powerful factors influencing the adoption (Tran Cao Uy et al., 2024 ; Sharma et al., 2025). UAV- and smartphone-based AI-based disease detection systems have been implemented in African contexts to deal with labour shortages and lack of agronomic knowledge especially with staple and horticultural crops. Benin and Ethiopian case studies demonstrate that deep learning models are effective to recognize fungal and viral diseases in the actual field conditions, which can help to protect crops and stabilize yield (Odounfa et al., 2025 ; Lankamo et al., 2025 ). However, there are still obstacles, such as data scarcity, affordability, trust, and institutional backing, that do not allow large-scale use (Mamabolo et al., 2025 ; Daum, 2025 ). Altogether, Asian and African studies highlight the idea that technological performance is not enough, effective implementation of AI-based crop disease detection has to be consistent with local agronomic solutions, inclusive design, and conducive policy and governance frameworks (High et al., 2025 ; Ahmad et al., 2025 ; Wu and Zhong, 2025 ). 4.2 AI-Based Yield Prediction and Resource Optimization ML/DL Models for Yield Forecasting The use of machine learning (ML) and deep learning (DL) has emerged as the focal point of crop yield prediction as they are able to cause complex, nonlinear interactions between climatic, soil, crop, and management variables. Conventional ML methods include random forests, support vector machines, gradient boosting, and k-nearest neighbors, which have been extensively used to predict the yields using historical yield data, meteorological, and soil attributes (Aijaz et al., 2025 ; Gupta and Kumar Pal, 2025 ). More recently, the predictive performance of DLs, such as artificial neural networks (ANNs), convolutional neural networks (CNNs), and recurrent neural networks (RNNs), has proven to be the best when working with high-dimensional data and/or spatiotemporal data (Ajith et al., 2025 ; Filippi et al., 2025 ). LSTM networks are becoming more popular in yield forecasting because they can recreate time-dependent correlations in weather and crop growth cycles and enhance the accuracy of the forecast in a variety of seasons (Mahale et al., 2024 ; Kuradusenge et al., 2024 ). The hybrid systems that involve satellite images, UAVs and IoT-based sensors, and ML/DL models augment the spatial resolution and real-time yield estimation further, as yield predictions will correspond to the goals of precision agriculture (Rashid et al., 2025 ; Sudha & Loret, 2026). Regardless of these developments, issues concerning the quality and transferability of data across agroecological areas and the interpretability of models of these models remain, which restrict a large-scale operational implementation (Filippi et al., 2025 ; Wu and Zhong, 2025 ). Agricultural and Nutrient Management Exactness. AI-assisted resource optimization is crucial to precision irrigation and nutrient management because the technology allows making site-specific and demand-driven decisions. Decision support systems based on MLs combine soil moisture sensors, weather forecasts, crop growth models, and estimations of evapotranspiration to automatically schedule irrigation to minimize water wastage and preserve or improve crop yields (Bayar et al., 2025 ; Adinarayana et al., 2024). Fuzzy logic methods and reinforcement learning have also been investigated in the field of adaptive control of irrigation in changing climatic conditions (Mana et al., 2024 ; Sudha and Loret, 2026). On the same note, AI-based nutrient management systems use ML algorithms to forecast nutrient deficiencies and suggest the best rates of fertilizer application depending on the soil characteristics, crop development, and yielding objectives (Mamabolo et al., 2025 ; Ajith et al., 2025 ). AIoT-based systems allow the real-time operation of nutrient delivery and close-loop control to improve nutrient utilization and reduce negative externalities of the environment in terms of leaching and greenhouse gases (Bayar et al., 2025 ; Gupta et al., 2025). Nevertheless, sensors are not completely reliable, cost calibration and poor digital infrastructure are major setbacks to the large scale adoption, especially in the small holder farming systems (Saiz et al., 2026 ; Daum, 2025 ). Data on Gains in Productivity and Efficiency of Input. According to the empirical observations of different agroecological settings, AI-based yield forecasting and improvement of resource optimization can result in quantifiable productivity increase and input efficiency enhancement. Research provides increasing yields due to decisions based on AI support in moderate to large increments, depending on the type of crops, availability of data, and size of implementation (Aijaz et al., 2025 ; Erickson and Fausti, 2021). ML-powered precision irrigation systems have continued to be shown as reducing water consumption without affecting yields, which leads to better water productivity and climate resilience (Bayar et al., 2025 ; Mana et al., 2024 ). Nutrient optimization made with the assistance of AI has also been linked to reduced fertilizer use and improved use of nutrients, which supports the goals of economic and environmental sustainability (Mamabolo et al., 2025 ; Ugwu et al., 2025 ). However, the sizes of achieved benefits greatly depend on the socio-economic factors, institutional backing, and the ability of farmers, to read and believe AI advice (Gardezi et al., 2023 ; Lee et al., 2024 ). To ensure a long-term productivity increase, the technological innovation needs to be accompanied by an inclusionary governance system, extension services, and moral deployment models (as highlighted in recent policy-oriented studies) (Omotayo et al., 2025 ; Ahmad et al., 2025 ; Wu and Zhong, 2025 ). 4.3 Sustainability and Environmental Outcomes Reduced Chemical Use Precision agriculture grounded in the use of artificial intelligence has been shown to have a lot of potential as far as the diminishing usage of chemical inputs like pesticides, herbicides, and synthetic fertilizers using the data-driven and precise interventions. The early detection of pest infestations, nutrient deficiencies, and crop stress allows the use of agrochemicals site-specifically and in time by AI-based crop monitoring systems based on machine learning, computer vision, and sensor networks (Majdalawieh et al., 2025 ; Padhiary et al., 2024 ). Providing the AI-based solution of decision support systems can reduce unnecessary chemical application and still ensure crop productivity by replacing blanket field-level application with variable-rate and spot-specific applications (Aijaz et al., 2025 ; Mamabolo et al., 2025 ). Disease detection based on the vision and AIoT platforms also help to reduce chemicals by allowing preventive and not reactive measures in crop protection (Bayar et al., 2025 ; Rashid et al., 2025 ). According to the empirical reviews, these strategies help to reduce the risks of environmental contamination and decrease the number of chemicals entering the adjacent ecosystems (Mana et al., 2024 ; Gupta and Kumar Pal, 2025 ). However, AI-guided reduction plans are successful depending on the sensor precision, algorithm and farmer confidence in the AI advice, which is not evenly distributed in the regions (Saiz et al., 2026 ; Gardezi et al., 2023 ). Water efficiency and Soil health. The technologies of AI-based precision agriculture have a direct effect on the enhanced water-use efficiency and better soil health through the optimization of irrigation timings, fertilizer application, and soil management methods. Soil moisture sensors, weather data sensors, and crop growth indicators together in machine learning models make it possible to make irrigation decisions based on real crop demand to avoid over-irrigation and soil degradation (Bayar et al., 2025 ; Adinarayana et al., 2024). These systems contribute to managing water sustainably because they reduce the amount of evapotranspiration losses and counteract the threat of salinization of intensive lands (Aijaz et al., 2025 ; Sudha & Loret, 2026). On the same note, AI-based systems to monitor soil health will support real-time evaluation of soil properties, allowing to balance nutrient levels in the soil and enhance soil retention of organic matter (Mamabolo et al., 2025 ; Ajith et al., 2025 ). AI applications can save the soil and prevent soil erosion by encouraging the efficient use of inputs, which helps to maintain the long-term soil productivity (Wu and Zhong, 2025 ). Nevertheless, the scalability of these advantages can be limited in smallholder-based systems because of socio-economic factors, such as access to digital infrastructure, high costs of sensor deployment, and similar aspects (Daum, 2025 ; Saiz et al., 2026 ). Climate-Smart Agriculture In response to the concept of climate-smart agriculture (CSA), AI-enabled agricultural systems are increasingly compatible with the principles of climate-smart agriculture (CSA) through boosting productivity, helping farmers to adapt to climate variability, and decreasing the emission of greenhouse gases. Sound AI climate resilience AI employs historical climatic data and real-time environmental provisions to make adaptive decisions in farm management in response to uncertain climatic conditions (Wu and Zhong, 2025 ; Vijayakumar et al., 2025). Accurate input control through AI helps to meet goals of mitigation by decreasing the energy-intensive consumption of fertilizers and decreasing emissions of wasteful irrigation and the use of chemicals (Mana et al., 2024 ; Omotayo et al., 2025 ). Additionally, the monitoring and predictive systems supported by AI contribute to CSA because they allow evidence-based design of policies and climate-related extension services (Hiywotu, 2025 ; High et al., 2025 ). Even within this alignment, scholars underline that the environmental sustainability of AI in agriculture is subject to inclusive governance systems, ethical applications and the incorporation of local knowledge to prevent the establishment of inequalities and technological lock, in (Ahmad et al., 2025 ; Atasoy, 2025 ). This means that the potential of AI in achieving complete climate-smartness cannot be achieved alone through technological, institutional, and socio-economic interventions (Wu and Zhong, 2025 ; Lankamo et al., 2025 ). 5. Farmer Perception and Adoption of AI Technologies 5.1 Awareness, Attitudes, and Behavioral Intentions Awareness, Attitudes, and Behavioral Intentions Digital literacy and AI Trust. Digital literacy is an important factor when determining awareness and trust of farmers in agricultural technologies based on artificial intelligence (AI). Few digital skills can make farmers unable to process AI-generated knowledge, which will result in a lack of trust in algorithm-based recommendations in fields like disease diagnosis, irrigation planning, and nutrient management (Daum, 2025 ; Aijaz et al., 2025 ). On the other hand, the increased digital competence allows farmers to be more certain about their data-driven decision-making and to use AI-enhanced tools of precision agriculture more knowledgeably (Wu and Zhong, 2025 ; Gupta and Kumar Pal, 2025 ). Transparency, data ownership and privacy are also factors that determine trust in AI systems. To gain trust in AI applications, the farmers are likely to be more convinced that systems are consistent with agronomic knowledge and give explanable recommendations instead of black box predictions (Gardezi et al., 2023 ; Omotayo et al., 2025 ). Research highlights the importance of socio-technical integration as a way to boost trust and promote the intention to adopt AI (when AI supplements the experience of farmers), as this concept can strengthen trust and foster positive intentions (Ryan et al., 2023 ; Ahmad et al., 2025 ). The participatory design and training interventions play a crucial role especially in the smallholder settings where trust and fear of AI technologies are likely to be built (High et al., 2025 ; Su et al., 2026 ). Perceived usefulness and ease of use. Attitudes of farmers to the adoption of AI technologies are largely shaped by the perceived usefulness. It is empirically proven that farmers tend to implement AI-based solutions in situations where tangible benefits such as better crop productivity, enhanced input utilization, early disease diagnosis, and minimized production risks are evident (Aijaz et al., 2025 ; Ugwu et al., 2025 ). Use of AI-based crop protection solutions, intelligent irrigation, and yield prediction have been demonstrated to be effective in improving the operational efficiency and accuracy of decisions, supporting positive attitude towards adoption (Bayar et al., 2025 ; Majdalawieh et al., 2025 ). Adoption intentions are also further moderated by ease of use mostly amongst farmers who have limited exposure to technology. The absence of contextual customization, demanding high levels of learning, and complex interfaces can demoralize adoptions despite perceived high usefulness (Özoğul, 2025 ; Saiz et al., 2026 ). Customized systems design, understandable user interfaces, and localized implementation of deployment plans would dramatically enhance the usability and adoption of AI technologies (Su et al., 2026 ; Sudha and Loret, 2026). These conclusions can be correlated with the views offered by the technology acceptance, where usefulness and ease of use have a combined effect on the behavioral intentions of farmers regarding AI-enabled precision agriculture (Sood et al., 2022 ; Lee et al., 2024 ). Extension services and peer networks. Agricultural extension services can be a key factor in determining the attitude and behavior awareness of farmers in terms of using AI technologies. The extension agents serve as intermediaries that convert the complex AI outputs into actionable and farm-level recommendations to lower cognitive and informational barriers to adoption (Lee et al., 2024 ; High et al., 2025 ). There are indicators that extension-based demonstrations, capacity-building practices, and field trial can have great impacts at raising farmers awareness and confidence in AI applications (Hiywotu, 2025 ; Vijayakumar et al., 2025). Peer networks also have the effect of social learning and diffusion of trust as a means of influence in adoption. The farmers are more likely to use AI technologies once they have seen positive results in other neighbors or even known people in the community (Nguyen et al., 2023 ; Tran Cao Uy et al., 2024 ). Positive attitudes and AI normalization among local agricultural systems are supported through informal peer interactions, cooperatives, and digitally integrated groups of farmers (Wu and Zhong, 2025 ; Lankamo et al., 2025 ). This interaction between the extension services and peer networks therefore forms an enabling ecosystem that facilitates the continuous and inclusive integration of AI technologies in agriculture. 5.2 Socio-Economic Determinants of Adoption Agricultural Area, Agricultural Revenue and Education. The size of farms is one of the most repeatedly found factors affecting the use of AI and precision agriculture technologies. Bigger farms are also more likely to take on AI-based solutions sooner because of their higher financial capabilities, economies of scale, and capacity to take risks related to high initial costs and uncertain payoff (Aijaz et al., 2025 ; Mana et al., 2024 ). According to the studies published in Smart Agricultural Technology and Journal of Agriculture and Food Research, medium- and large-scale commercial farms are more likely to use AI-enabled irrigation systems, disease detection systems, and decision-support tools than smallholders (Bayar et al., 2025 ; Mamabolo et al., 2025 ). Farm size is closely interacting with income level, determining the access to AI infrastructure and the long-term sustainability of its use. Farmers with higher incomes have a chance to invest in sensors, AIoT systems, UAVs, and analytics services based on a subscription (Rashid et al., 2025 ; Omotayo et al., 2025 ). In the low-income situation, on the other hand, smallholder farmers tend to be credit constrained, have less access to digital infrastructure, and high opportunity costs, which slack adoption despite any gain in productivity (Saiz et al., 2026 ; Sharma et al., 2025). Digital skills and education come in as a huge mediator of how the availability of technology is related to actual use. It is empirically tested and reviewed that farmers having a higher level of formal education and having been introduced to digital tools earlier have a stronger perceptual attitude towards usefulness and ease of use, which increases faster AI adoption decisions (Daum, 2025 ; Nguyen et al., 2023 ). In Global South, there is still low digital literacy as a structural obstacle, especially in cases where AI systems use sophisticated interfaces or ability to process data (Wu and Zhong, 2025 ; Tran Cao Uy et al., 2024 ). Gender and Generation Dimensions. The issue of gender inequality is vital in the process of influencing the patterns of adoption of AI in the agricultural sector. Women farmers, particularly in developing areas are frequently unequally supplied with land, credit, training, and digital devices, limiting their involvement with AI-facilitated agriculture (Dittmer et al., 2025 ; Lankamo et al., 2025 ). Research highlights that women are restricted in their access to digital extension programs and technology tests by socio-cultural norms and institutes even in instances when AI tools are technically available (High et al., 2025 ; Ahmad et al., 2025 ). The differences between generations also affect the adoption. Farmers of younger age are more willing to use AI, data-driven decision-making, and automation, which is explained by being more digitally familiar and risk-takers (Sood et al., 2022 ; Gardezi et al., 2023 ). Conversely, more senior farmers tend to be guided by experience and can view AI systems as intricate or incompatible with the existing ways of doing things, which lowers the intention to adopt (Ozoglu, 2025). Nonetheless, it is believed that intergenerational learning and peer influence may help to reduce them, at least, when AI tools are created in the framework of human-centered and participatory design (Su et al., 2026 ; Lee et al., 2024 ). Regional Inequality and the Global South Regional inequalities are still acute in the spread of AI technologies, as their usage is not only concentrated in countries with high income but also in developed agricultural areas. Structural issues, including poor digital connectivity, poor signal connectivity, high sensor prices, and divided land parcels are among the major barriers to adoption in the Global South (Vijayakumar et al., 2025; Abdulraheem et al., 2026 ). African and South Asian studies indicate that although AI applications are no longer seen as a limited pilot project with questionable possibilities, many of them are still in the pilot phase (Mamabolo et al., 2025 ; Odounfa et al., 2025 ). The nature of AI technologies adapted is also determined by socio-economic inequalities across regions. Mobile-based advisory systems and low-cost decision-support tools are more likely to receive adoption in low-and-middle-income countries compared to capital-intensive automation and robotics (Nguyen et al., 2023 ; High et al., 2025 ). Moreover, the policy gaps, ineffective extension, and restrained institutional assistance contribute to the regional disparities, suggesting the necessity of the inclusive AI governance and region-specific deployment strategies (Omotayo et al., 2025 ; Atasoy, 2025 ). On the whole, the literature tends to agree that AI implementation in agriculture is not a technological process but a profoundly socio-economic phenomenon, which is determined by the characteristics of farms, social systems, and developmental trends. These determinants will have to be addressed when it comes to providing equitable and sustainable AI-driven agricultural transformation, especially in the Global South (Wu and Zhong, 2025 ; Ryan et al., 2023 ). 5.3 Human-Centered AI and User Experience Human-centered artificial intelligence (HCAI) has been proposed as a more important paradigm to promote the acceptance, performance, and viability of AI technologies in agriculture. In addition to its technical performance, the trust, engagement, and sustainability of farmers with AI-driven systems are influenced by user experience factors including usability, transparency, cultural relevance, and accessibility (Wu and Zhong, 2025 ; Omotayo et al., 2025 ). The literature on AI solutions reviewed by Scopus in the recent past has a growing focus on the need to design AI solutions, basing them on the needs, situations, and abilities of farmers to facilitate a comprehensive and equitable transformation of agriculture. Co-Design and Participatory Development. The extensive use of co-design and participatory development methods is also accepted as a key to the harmonization of AI technologies and real-world farming. Instead of viewing farmers as passive end-users, participatory models engage them during the design, testing, and refinement phases of AI systems, which guarantees that the tools are designed to capture local agronomic knowledge and decision-making (Ryan et al., 2023 ; Mana et al., 2024 ). Precision agriculture and AIoT application evidence indicates that participatory engagement can increase system relevance, lower technological change resistance, and perceived usefulness among farmers (Bayar et al., 2025 ; Mamabolo et al., 2025 ). The human-centered design has especially been found to work in smallholder and resource-constrained environments, where the standardized AI solutions fail to capture the variety of cropping systems, the patchwork nature of landholdings, and socio-economic limits. Research points out that participatory pilots and feedback loops enable adapting the AI-based disease detectors, irrigation scheduling, and advisory systems to the local conditions, thus enhancing adoption results (Su et al., 2026 ; High et al., 2025 ). Besides, mediating the participation of an AI in development wherein technical concepts are transformed into practice-based solutions is the role of extension agents and farmer organizations (Lee et al., 2024 ; Tran Cao Uy et al., 2024 ). Explainable AI and Localized Interfaces. Explainable AI (XAI) is regarded more and more as a foundation of trust and acceptance of agricultural AI systems by users. Complex black-box models are highly accurate, but with little transparency and interpretability, they may not be able to win the trust of farmers (Gardezi et al., 2023 ; Majdalawieh et al., 2025 ). The literature also highlights that farmers will tend to trust AI recommendations more when the systems have understandable explanations of the outputs, i.e., visual indicators, rule-based logic, or simplified confidence scores (Aijaz et al., 2025 ; Wu and Zhong, 2025 ). Localized user interfaces also improve usability by tailoring AI outputs in accordance with the cognitive frames and realities in the work of farmers. Demonstrated to be more efficient as far as the decision-making process is concerned and to create less cognitive load, context-specific dashboards, mobile-based notifications, and visual cues have been adapted to local crops and practices (Padhiary et al., 2024 ; Rashid et al., 2025 ). Research also mentions that explainability is especially significant in high-stakes decisions, like the choice of pesticides or the time of irrigation, where farmers require to know the reasoning behind AI-based recommendations in order to reduce the perceived risks (Omotayo et al., 2025 ; Özoğul, 2025 ). Accessibility, Cultural Relevancy, and Language. The issue of access to AI technologies in the Global South revolves around language and cultural relevance. Studies always show that AI systems optimized on the main languages of dominance or belonging to the global community impose an obstacle to farmers with low levels of formal education or literacy (Daum, 2025 ; Dittmer et al., 2025 ). Voice-based advisory systems, local-language interfaces, and icon-based visualizations are much more effective in understanding and interacting with the application, especially among smallholder farmers (Nguyen et al., 2023 ; Sharma et al., 2025). Cultural relevance goes beyond language and encompasses cultural correspondence to local agricultural practices, perceptions of risk and social constructs. Research stresses that the technological savvy of the AI tools lacks acceptance in many instances when indigenous knowledge systems or even known practices are ignored (Atasoy, 2025 ; Hiywotu, 2025 ). On the other hand, culturally adaptive AI systems, i.e., systems that incorporate local heuristics and knowledge of farmers, are less likely to be perceived as disruptive and have more chances of being perceived as supportive (High et al., 2025 ; Lankamo et al., 2025 ). The accessibility also includes affordability, compatibility with the devices, and infrastructure limitations. Smaller AI models, offline experience and mobile-first designs are also being encouraged in order to make sure that they can be used in areas with weak connection and lack of hardware access (Saiz et al., 2026 ; Abdulraheem et al., 2026 ). Collectively, these anthropocentric considerations highlight the fact that the successful implementation of AI in agriculture is not only a factor of the functionality of the algorithms but also the degree of intelligibility, applicability, and accessibility of the systems to different farming groups. 6. Barriers and Enablers for AI Implementation 6.1 Technical and Infrastructure Barriers Although artificial intelligence (AI) applications are increasingly becoming mature in agriculture, its mass application to the enterprise is still limited due to the ongoing technical and infrastructural challenges. Such challenges are especially acute regarding developing agricultural systems and smallholder agriculture, in which digital ecosystems have remained fragmented and unevenly developed. Connectivity Gaps Dependable online connectivity is a pre-requisite of AI-supported farming technologies, such as cloud-based decision support systems, AIoT systems, UAV analytics as well as real time sensor networks. Nevertheless, there are still significant connectivity differences between urban and rural areas, which restricts the functionality and scalability of precision farming through AI. Poor broadband connectivity, unreliable mobile signals, and excessively expensive data transfer create barriers in real-time data derived and model implementation, particularly applications like smart irrigation, illness overview, and self-operated machinery (Bayar et al., 2025 ; Rashid et al., 2025 ). Research articles are consistent in reporting that lack of quality digital infrastructure limits access to AI services by farmers and reduces feedback loops needed by adaptive learning systems (Saiz et al., 2026 ; Ishore et al., 2025 ). Such gaps increase the gap in digital inequalities and diminish the benefits in productivity and sustainability that AI-based innovations in agriculture promise (Wu and Zhong, 2025 ; Ahmad et al., 2025 ). Data Sparingness and Data Quality. The basis of AI systems in agriculture is data-intensive, as large amounts of high-quality, context-dependent data are needed to train, validate and continue improving. Nevertheless, the lack of data is still a severe bottleneck especially in places where smallholder farming is predominant and there are mixed agro-ecological contexts. The weaknesses of AI models include limited access to labeled datasets, uneven data collection, temporal and spatial gaps in data, which deteriorate the strength and generalizability of AI models (Aijaz et al., 2025 ; Filippi et al., 2025 ). Besides, the quality of data, including sensor noise, missing data, biased sample, and non-standardization, lowers the predictive accuracy and confidence of farmers in AI results (Majdalawieh et al., 2025 ; Gupta and Kumar Pal, 2025 ). A number of surveys point out that the lack of integration of local agronomic insights and ground-truth validation further limits the model transferability between regions and cropping systems (Mana et al., 2024 ; Upadhyay et al., 2025 ). To overcome these issues, it is necessary to invest in data governance systems, participatory data collection, and interoperable agricultural data standards. Interoperability Challenges The other major technical obstacle to successful AI implementation in agriculture is interoperability. The existing state of the digital agriculture sector can be described as the increase in the number of proprietary platforms, heterogeneous sensors, and vendor-specific software ecosystems that are not always compatible. Consequently, the combination of data streams of IoT devices, UAVs, farm machinery, and external data is still complicated and expensive (Padhiary et al., 2024 ; Rashid et al., 2025 ). The lack of interoperability hinders the transfer of data and the creation of farm-scale decision support devices (Omotayo et al., 2025 ; Ryan et al., 2023 ). Moreover, the fact that there are no common principles of data formats, communication protocols and the implementation of the AI models complicates the scalability of the systems and their long-term sustainability (Özoğul, 2025 ; Abdulraheem et al., 2026 ). Researchers point out that interoperability barriers will only be overcome through a concerted effort by technology vendors, policy makers and research institutions to ensure open architectures, standardization and modular system design (Wu and Zhong, 2025 ; Vijayakumar et al., 2025). 6.2 Institutional and Economic Barriers. Although artificial intelligence (AI) technologies have a high potential of increasing productivity, sustainability and risks management in agriculture, economic and institutional barriers are also major limiting factors to their diffusion. The smallholder and resource-constrained farmers are disproportionately impacted by these barriers and result in unequal access to AI-enabled innovations and slows system-wide change. Cost of AI Tools The prohibitive initial and maintenance expenses that are involved in AI-agricultural technologies continue to be one of the biggest obstacles to adoption. Small and medium-scale farmers may not be financially able to invest in sensors, UAVs, smart machinery, cloud subscriptions, and data analytics platforms. Besides the purchase of the hardware, the maintenance of the system, updates of the software, data storage, technical support, among others, make this total cost of ownership even greater (Aijaz et al., 2025 ; Padhiary et al., 2024 ). Empirical and review research has shown that perceived economic risk and uncertain payoff on investment deter farmers to use AI solutions, which is more evident when prices are volatile and uncertain about the climate (Mamabolo et al., 2025 ; Özoğul, 2025 ). Even though AIoT and automation technologies have the potential to achieve long-term efficiency benefits, their cost is firmly linked to the economies of scale, which supports asymmetries in adoption between large commercial farms and smallholders (Bayar et al., 2025 ; Gupta and Kumar Pal, 2025 ). Absence of Insurance and Credit Intersection. The minor integration of the AI technologies into the agricultural credit and insurance frameworks is a severe institutional chokepoint. Although AI-controlled analytics has potential, the ability to enhance risk evaluation, yield forecasting, and loss verification, they are not often offered in formal financial services available to farmers (Omotayo et al., 2025 ; Wu and Zhong, 2025 ). Most organizations do not provide new types of credit products, which acknowledge digital assets or data-based performance measurements, limiting farmers when making AI investments (Sood et al., 2022 ; Hiywotu, 2025 ). Equally, the lack of AI-based crop insurances schemes lowers the use of technology incentives because farmers will not be under the protection of sufficient safety nets against climatic risks and market risks (Saiz et al., 2026 ; Vijayakumar et al., 2025). Researchers stress that to reduce the risks of adoption and increase economic feasibility of precision agriculture, there is a need to have greater alignment between AI systems, financial institutions, and insurance companies (Ryan et al., 2023 ; Ahmad et al., 2025 ). Disjointed Policy Ecosystems. Scattered and inconsistent policy landscapes also restrict the scalability of AI in agriculture. Digital agriculture strategies in most countries are spread over various ministries and agencies, which leads to overlapping mandates, regulatory confusion, and low coordination between policies of innovation, data governance, and rural development (Ahmad et al., 2025 ; Omotayo et al., 2025 ). The absence of explicit policies on data possession, privacy, interoperability and deployment ethics of AI poses uncertainties to both the technology providers and the end-users (Atasoy, 2025 ; Özoğul, 2025 ). In addition, the lack of governmental funding of digital public goods, including open datasets, advisory services, and AI services that are based on extensions, decreases the institutional support of inclusion in adoption (Daum, 2025 ; Dittmer et al., 2025 ). They indicate that coherent policy frameworks, as well as incentive-based programs and public- private partnerships, have a key role in building trust, lowering transaction costs and facilitating sustainable integration of AI technologies in agricultural value chains (Wu and Zhong, 2025 ; Vijayakumar et al., 2025). 6.3 Enablers and Success Factors A combination of institutional, technological, and policy-related enablers is significant in the success of artificial intelligence (AI) adoption and scaling in agriculture. Recent scopus-indexed literature points out that in addition to technological preparedness, models of service delivery, collaborative governance, digital advisory systems, and enabling public policies have a decisive role to play in the determination of impact and sustainability. AI-as-a-Service Models AI-as-a-Service has become an important facilitator of reduced barriers of entry to advanced digital technologies in agriculture. Rather than demanding farmers to invest in expensive hardware, software, and skilled employees, AIaaS enables the utilization of analytics and decision-support applications and predictive models via cloud-based technologies on subscription or a pay-per-use basis. Research notes that AIaaS allows deploying precise farming solutions (including yield prediction, disease detection, and optimization of irrigation) on a larger scale especially to small and medium-scale farmers (Aijaz et al., 2025 ; Mana et al., 2024 ). AIaaS is also highly associated with AIoT systems, in which sensor data of the fields, UAVs, and smart equipment are centrally processed to show real-time insights, thus increasing the efficiency of the resources and operational decision-making (Bayar et al., 2025 ; Rashid et al., 2025 ). In addition, service-oriented AI provision eliminates risks associated with system maintenance, model updates, and data security, which tend to be mentioned as obstacles to the use of AI on the farm (Omotayo et al., 2025 ; Wu and Zhong, 2025 ). Public-Private Partnerships (PPP). The concept of public-private partnerships is commonly identified as potential success factors in fast-tracking the process of AI innovation and diffusion into the agricultural sector. PPPs help leverage the strengths of data analytics, platform development, and commercialization (that are part of the capabilities of the private sector) with the objectives of food security, sustainability, and inclusion (that are part of the goals of the public sector) (Ahmad et al., 2025 ; Ryan et al., 2023 ). According to evidence PPPs are an important driver of the piloting of AI solutions, the development of a common data infrastructure, and alignment to local agronomic and socio-economic environments (Majdalawieh et al., 2025 ; Gupta and Kumar Pal, 2025 ). Moreover, PPPs can be used to build trust in farmers by integrating AI tools into publicly approved programs and extension systems and, therefore, decrease farmers' distrust of individual digital technologies (Gardezi et al., 2023 ; Omotayo et al., 2025 ). Digital Extension Systems Digital extension systems constitute a radical facilitator of reaching out to farmers via AI technologies. AI-based advisory systems, mobile apps, and decision-support applications can help reach more customers with its traditional extension services and provide customized recommendations in real-time, location-based, and personalized (High et al., 2025 ; Tran Cao Uy et al., 2024 ). Literature highlights the importance of digital extension systems in increasing the ability of farmers to understand AI outputs and enhancing perceived usefulness and intention to adopt AI-driven practices (Lee et al., 2024 ; Wu and Zhong, 2025 ). Combining AI and digital extension also facilitates continual learning based on feedback loops, peer interaction, and data-driven demonstrations that are especially useful in smallholder and resource-limited settings (Dittmer et al., 2025 ; Saiz et al., 2026 ). Open Data Platforms and Government Subsidies. The essential success factors for the equitable use of AI in agriculture are government assistance, such as subsidies and open data programs. Digital infrastructure, smart sensors, and AI-enabled machines subsidies lower financial risks and encourage early adoption, particularly in smallholder farmers (Hiywotu, 2025 ; Vijayakumar et al., 2025). Agri-data websites such as weather, soil, crop conditions, and market data are also recurrently mentioned as competition-free inputs to train powerful AI models and develop innovation ecosystems (Wu and Zhong, 2025 ; Atasoy, 2025 ). Both transparency and interoperability and innovation are facilitated by publicly available datasets through allowing startups, researchers, and extension agencies to develop AI solutions more specific to local needs (Aijaz et al., 2025 ; Özoğul, 2025 ). Combined, subsidies and open data policies can empower the facilitation of sustainable, inclusive, and ethically based AI implementation in the agricultural sector (Omotayo et al., 2025 ). 7. An Integrated Framework for AI Adoption in Smallholder Agriculture 7.1 Smallholder-Centric AI Adoption Framework I. Technology Layer: Accuracy, Scalability, and Explainability. Technology Layer smallholder agriculture AI adoption focuses on AI tool and model accuracy, scalability and explainability. The precision makes smallholders with limited resources reliable in crop prediction, detection of pests, and optimization of irrigation, which is essential (Aijaz et al., 2025 ; Majdalawieh et al., 2025 ). Scalability enables the application of AI-driven IoT devices, UAVs, and AIoT solutions to different scales of farms to make more farmers use the precision farming practices (Bayar et al., 2025 ; Rashid et al., 2025 ). Explainability would allow farmers to be familiar with AI recommendations and more likely to trust and implement the tools, overcoming the challenges caused by black-box AI systems (Wu and Zhong, 2025 ; Mana et al., 2024 ). Combined, these technological factors form the basis for effective and acceptable AI systems for smallholder farmers. II. Human Layer: Perceived Value, Trust, and Skills. The Human Layer pays attention to the socio-cognitive factors that affect the adoption of AI. The confidence in AI systems is a decisive factor because smallholders will be more willing to use solutions that can be seen as reliable and transparent (Gardezi et al., 2023 ; Su et al., 2026 ). To use, interpret, and gain access to AI technologies, one will require sufficient skills, such as digital literacy and knowledge of AI-assisted farming practices (Daum, 2025 ; High et al., 2025 ). Last but not least, the perceived value of AI systems, the perception of farmers about the positive effect of the systems on productivity, income, or sustainability, is a strong determinant of adoption (Nguyen et al., 2023 ; Sood et al., 2022 ). Any initiatives that address these elements will ensure more involvement and higher chances of the continued adoption of AI among smallholders. III. System Layer: Policy, infrastructure, and Markets. System Layer is an enabling environment to adopt AI. Policies on the national and regional levels, including subsidies, assistance in the development of innovations, and regulation, become favorable to the implementation of AI (Omotayo et al., 2025 ; Ahmad et al., 2025 ). The implementation of AI technologies on a large scale and inclusion of the smallholders requires adequate infrastructure, such as power, internet access, and digital services (Saiz et al., 2026 ; Tran Cao Uy et al., 2024 ). Lastly, economic incentives to smallholders to use AI are reinforced by the presence of market mechanisms that offer fair prices, access to inputs, and trading platforms of selling outputs (Hiywotu, 2025 ; Erickson and Fausti, 2021). The combination of these systemic components will provide the adequate support of technological and human capabilities of the agricultural ecosystem. 7.2 AI Innovation to Productivity Gains Pathways. The channels by which AI innovations can be converted into productivity increase are mediated by adoption processes and cyclic engagement between smallholder farmers and technology developers. Mediating AI Impact through Adoption. Adoption is an important intermediary between AI technologies and real gains in agricultural productivity. Although AI-based technologies, including precision irrigation, pest detection, and yield predictions, have a high technical potential, their application will rely on the adoption by smallholders (Aijaz et al., 2025 ; Bayar et al., 2025 ). The willingness of farmers to use AI systems, perceptions of usefulness, and competency directly affect the extent of AI interventions generating demonstrable results in increase in crop yields, resource management, and revenue (Nguyen et al., 2023 ; Sood et al., 2022 ). Moreover, it depends on the availability and convenience of technology, such as smartphone applications, IoT sensors, and automated farm machines (Padhiary et al., 2024 ; Rashid et al., 2025 ). Feedback between Technology Developers and Farmers. The proper implementation of AI depends on the feedback loops that would enable further learning between farmers and developers. Smallholders can deliver current information and contextual data regarding the state of the soil, pest behavior, and crop data that can be used to optimize machine learning models and enhance AI decision-making (Wu and Zhong, 2025 ; Upadhyay et al., 2025 ). Human-oriented AI tools and participatory design practices are used to make sure that the technologies can adapt to the needs of farmers, and the cycle of innovation-adoption-productivity improvement develops (Su et al., 2026 ; High et al., 2025 ). Such iterative feedback mechanisms prove especially useful when dealing with resource-constrained settings where generic AI solutions are prone to fail when they do not adapt locally (Omotayo et al., 2025 ; Mana et al., 2024 ). 7.3 Comparison to Existing Models. The Framework Extension of Agriculture 4.0 / 5.0 Models. Agriculture 4.0 and 5.0 models focus on automation, IoT connectivity, and AI-driven optimization (Baladraf et al., 2025 ; Gupta et al., 2025), whereas the Smallholder-Centric AI Adoption Framework directly incorporates technological, human, and system layers with the view of considering the smallholder-specific constraints. In contrast to the generic smart agriculture models, this framework emphasizes the importance of the element of trust, skills, and perceived value in adoption, as well as facilitating policies, infrastructure, and market support (Tran Cao Uy et al., 2024 ; Daum, 2025 ). This is in the holistic approach such that the AI innovations are not only of an advanced technical nature, but also affordable and socially accessible to the small scale farmers. Value Addition to the Developing-Country Situations. The framework also brings considerable value to the context of the developing countries by considering socio-technical and economic challenges that restrict the adoption of AI. It also focuses on participatory design, integration of local knowledge and feedback loops to make AI interventions specific to local crop systems, resource availability, and market structures (Hiywotu, 2025 ; Erickson and Fausti, 2021). Also the framework takes into consideration policy and infrastructure gaps that are typical of low- and middle-income countries, building a more robust and inclusive channel between the initiation of AI innovations and productivity improvements (Omotayo et al., 2025 ; Ahmad et al., 2025 ). In general, it offers a viable blueprint of how to transform the latest AI technologies into viable productivity and income gains to the smallholders. 8. Implications of Policy, Design, and Research. The introduction of AI in smallholder agriculture must be a collective effort in policy, technology development, and research priorities. Stakeholders can maximize the benefits of AI to achieve sustainable agricultural productivity by solving socio-technical obstacles and providing inclusivity. 8.1 Policy Implications. Digital Infrastructure Investment: A successful application of AI depends on the good digital infrastructure, such as access to reliable internet connectivity, wireless networks, and mobile platforms, which are typically lacking in rural areas of the developing world (Tran Cao Uy et al., 2024 ; Daum, 2025 ). Digital infrastructure investments make it possible to collect data in time, use it in precision farming, and provide better access to the market to smallholders (Ahmad et al., 2025 ; Wu and Zhong, 2025 ). Inclusive AI Governance: Policymakers should come up with governance systems that are inclusive to all stakeholders and safeguard the smallholder against any risks, such as misuse of data and algorithm bias (Omotayo et al., 2025 ; Abdulraheem et al., 2026 ). Inclusive AI governance leads to accountability, transparency and local stakeholder involvement in AI policy making. Smallholder-Centric AI Solutions: National and regional solutions must target smallholder farmers through subsidies of AI-enabled solutions, extension, and combining AI solutions with local agricultural knowledge systems (Hiywotu, 2025 ; High et al., 2025 ). Specific policies will help to close the technology gap and help to increase productivity by providing benefits to resource-constrained farmers. 8.2 Technology Developers Implications. Design to Low-Resource Environments: AI systems must be designed to run well in low-resource environments, being intermittently connected, with limited access to hardware, and a variety of cropping conditions (Padhiary et al., 2024 ; Bayar et al., 2025 ). Mobile applications and smallweight AI tools and offline capabilities are necessary to make them usable by smallholders. Reliable and understandable AI: Trust is a significant factor in the use of AI. Explainable AI models should be a priority among developers that give unambiguous, practical advice and allow farmers to interpret results of the algorithms (Gardezi et al., 2023 ; Su et al., 2026 ). Open AI systems will lessen non-cooperation and build more acceptance among smallholders. Co-Creation with Farmers: Participatory design solutions, in which the farm is included in the process of creating the AI, enhance relevance, usability, and adoption rates (High et al., 2025 ; Mana et al., 2024 ). Developers and farmers can also provide feedback on AI tools to refine them over time through feedback loops to align with local agricultural practices and social-economic settings (Wu and Zhong, 2025 ; Upadhyay and others, 2025). 8.3 Future Research Implications. Longitudinal Impact Studies: Longitudinal studies that could evaluate long-term impacts of AI adoption on productivity, income, and sustainability are required (Aijaz et al., 2025 ; Erickson and Fausti, 2021). Knowledge of temporal impacts will help policymakers and developers to come up with more efficient interventions. Gender and Equity Dimensions: To make sure that AI adoption advantages all demographic groups, such as women farmers and marginalized communities, future studies are required to examine the impact of AI adoption on the specified groups (Nguyen et al., 2023 ; Sood et al., 2022 ). AI Ethics in Rural Areas: Ethical concerns, such as data privacy, algorithmic fairness and informed consent have not been studied in rural agricultural areas (Omotayo et al., 2025 ; Abdulraheem et al., 2026 ). Studies ought to examine the culturally appropriate systems to protect the smallholders and at the same time enhance AI development. 9. Conclusion The paper describes the importance of artificial intelligence in terms of improving productivity, sustainability, and resilience in smallholder farming. The offered framework unites technological, human, and systemic aspects, which contributes to a full-scale explanation of the ways of AI adoption and its possible consequences. Summary of Key Findings: The study has established that the implementation of AI is not only contingent upon technological, i.e., accuracy, scalability, and explainability, but also upon human factors, i.e. trust, skills and perceived value, and system-level enablers, i.e. policy, infrastructure and market access (Aijaz et al., 2025 ; Bayar et al., 2025 ; Wu and Zhong, 2025 ). The results show that feedback loops exist between farmers and technology developers in the mediation of adoption, and co-creation and participatory design are the most important elements of successful adoption (High et al., 2025 ; Su et al., 2026 ). Restatement of Research Objectives: The paper has approached the research goals of comprehending the adoption of AI through the perspective of smallholders, visualizing the routes to productivity improvement, and placing the paradigm in the context of current Agriculture 4.0 and 5.0 paradigms. The findings reveal that the suggested framework expands existing models by including socio-technical and equity aspects, which are especially critical in the framework of developing countries where resource-related limitations and digital divides still exist (Hiywotu, 2025 ; Tran Cao Uy et al., 2024 ). The stress on Human-Centered, System-Aware AI: The research confirms that AI cannot be a completely technical solution. To make sure that it will be meaningfully adopted and used, it is imperative to incorporate human-centered design, explainability, and trust-building. Similarly, system-sensitive interventions such as enabling policies, market connections, and infrastructure are needed to implement AI inventions as actual productivity (Omotayo et al., 2025 ; Ahmad et al., 2025 ; Mana et al., 2024 ). Placing AI as an Enabler, not a Standalone Solution: AI should be regarded as an enabler of smallholder productivity, and not a solution in and of itself. It relies on the incorporation of local knowledge, the involvement of farmers, and institutional support. AI can be used in a targeted way to change the smallholder agriculture sector through achieving sustainability, resilience, and equitable growth (Erickson and Fausti, 2021; Abdulraheem et al., 2026 ; Padhiary et al., 2024 ). Declarations Funding This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Clinical Trial Number Clinical trial number: not applicable. Ethics, Consent to Participate, and Consent to Publish Ethics, consent to participate, and consent to publish: not applicable. This study does not involve human participants or animals and therefore does not require ethical approval or consent. References Abdulraheem, M. I., Khan, Z. H., & Hu, J., et al. (2026). AgriTech: Prospects and challenges of robotics, artificial intelligence & 5G technologies in agriculture. Journal of Ambient Intelligence and Smart Environments. https://doi.org/10.1177/18761364251409357 Ahmad, A., Murad, M., Utra, U., Yussof, N. S., Fazial, F. F., & Lazim, A. (2025). Transforming food systems with artificial intelligence: Challenges, governance, and pathways for sustainable integration. Food Reviews International. https://doi.org/10.1080/87559129.2025.2585134 Ajith, S., Vijayakumar, S., & Elakkiya, N. (2025). Yield prediction, pest and disease diagnosis, soil fertility mapping, precision irrigation scheduling, and food quality assessment using machine learning and deep learning algorithms. Discover Food, 5, 67. https://doi.org/10.1007/s44187-025-00338-1 Aijaz, N., Lan, H., Raza, T., Yaqub, M., Iqbal, R., & Pathan, M. S. (2025). Artificial intelligence in agriculture: Advancing crop productivity and sustainability. Journal of Agriculture and Food Research, 20, 101762. https://doi.org/10.1016/j.jafr.2025.101762 Almazmomi, N. K. (2025). Artificial intelligence-driven blockchain and Internet of Things framework for secure data management in precision agriculture. Journal of High Speed Networks, 31 (3). https://doi.org/10.1177/09266801251331816 Atasoy, Y. (2025). Agriculture by algorithm: Big data, digitalization, and biotechnology under climate change. Science, Technology, & Human Values. https://doi.org/10.1177/01622439251321233 Baladraf, T. T., Marimin, Hermadi, I., & Ismayana, A. (2025). Review of Agriculture 5.0: Artificial intelligence, IoT, and machine learning. Journal of Agricultural & Food Information. https://doi.org/10.1080/10496505.2025.2583361 Bayar, J., Ali, N., Cao, Z., Ren, Y., & Dong, Y. (2025). Artificial intelligence of things (AIoT) for precision agriculture: Applications in smart irrigation, nutrient and disease management. Smart Agricultural Technology, 12, 101629. https://doi.org/10.1016/j.atech.2025.101629 Cardak, M. C., Golec, M., & Gill, S. S. (2025). FarmTechAI: Artificial intelligence-based modern farmer management system. In A. Kumar, J. P. Verma, & R. Jain (Eds.), Emerging smart agricultural practices using artificial intelligence (Chap. 12). Wiley. https://doi.org/10.1002/9781394274277.ch12 Clay, D. E., Brugler, S., & Joshi, B. (2024). Will artificial intelligence and machine learning change agriculture: A special issue. Agronomy Journal, 116 (3). https://doi.org/10.1002/agj2.21555 Daum, T. (2025). Digitalization and skills in agriculture. Outlook on Agriculture, 54 (2). https://doi.org/10.1177/00307270251336474 Dittmer, K. M., Burns, S., & Wollenberg, E. (2025). Digital tool innovations for smallholder inclusion. Outlook on Agriculture, 54 (3). https://doi.org/10.1177/00307270251331644 Filippi, P., Han, S. Y., & Bishop, T. F. A. (2025). On crop yield modelling, predicting, and forecasting and addressing the common issues in published studies. Precision Agriculture, 26, 8. https://doi.org/10.1007/s11119-024-10212-2 Gardezi, M., Joshi, B., Rizzo, D. M., Ryan, M., Prutzer, E., Brugler, S., & Dadkhah, A. (2023). Artificial intelligence in farming: Challenges and opportunities for building trust. Agronomy Journal, 115 (4). https://doi.org/10.1002/agj2.21353 Gupta, G., & Kumar Pal, S. (2025). Applications of artificial intelligence in precision agriculture. Discover Agriculture, 3, 61. https://doi.org/10.1007/s44279-025-00220-9 Hiywotu, A. M. (2025). Advancing sustainable agriculture for goal 2: Zero hunger – A comprehensive overview of practices, policies, and technologies. Agroecology and Sustainable Food Systems, 49 (7), 1027–1055. https://doi.org/10.1080/21683565.2025.2451344 High, C., Singh, N., & Nemes, G. (2025). Artificial intelligence for agricultural extension: Supporting transformative learning among smallholder farmers. Journal of Development Policy and Practice, 11 (1). https://doi.org/10.1177/2455133325134522 Ishore, B., Bhargav, A. M., Patel, S. K., Kumar, V., Sinha, J., & Kumar, S. (2025). Advancing smart agriculture: How IoT and 5G technologies enhance precision farming and sustainability. International Journal of Communication Systems. https://doi.org/10.1002/dac.70310 Kuradusenge, M., et al. (2024). SMART-CYPS: An intelligent IoT and machine learning–powered crop yield prediction system for food security. Discover Internet of Things, 4, 23. https://doi.org/10.1007/s43926-024-00079-0 Lankamo, A. A., Ramalingam, D., & Dira, S. J., et al. (2025). Adoption dynamics of climate-smart agriculture. SAGE Open. https://doi.org/10.1177/21582440251343674 Lee, C.-L., Strong, R., Briers, G., Murphrey, T., Rajan, N., & Rampold, S. (2024). Factors predicting innovation-decisions: The effects of performance expectancy, social influence, and facilitating conditions on U.S. Extension’s promotion of precision agriculture technologies. NJAS: Impact in Agricultural and Life Sciences, 96 (1). https://doi.org/10.1080/27685241.2024.2420111 Mahale, Y., et al. (2024). Crop recommendation and forecasting system for Maharashtra using machine learning with LSTM. Discover Sustainability, 5, 134. https://doi.org/10.1007/s43621-024-00292-5 Mamabolo, E., Mashala, M. J., Mugari, E., Mogale, T. E., Mathebula, N., Mabitsela, K., & Ayisi, K. K. (2025). Application of precision agriculture technologies for crop protection and soil health. Smart Agricultural Technology, 12, 101270. https://doi.org/10.1016/j.atech.2025.101270 Majdalawieh, M., Martins, C., Radi, M., Alaraj, M., & Khan, S. (2025). Precision agriculture in the age of AI: A systematic review of machine learning methods for crop disease detection. Smart Agricultural Technology, 12, 101491. https://doi.org/10.1016/j.atech.2025.101491 Mana, A. A., Allouhi, A., Hamrani, A., Rehman, S., El Jamaoui, I., & Jayachandran, K. (2024). Sustainable AI-based production agriculture: Exploring AI applications and implications in agricultural practices. Smart Agricultural Technology, 7, 100416. https://doi.org/10.1016/j.atech.2024.100416 Minhans, K., Sharma, S., Sheikh, I., Alhewairini, S. S., & Sayyed, R. (2025). Artificial intelligence and plant disease management: An agro-innovative approach. Journal of Phytopathology. https://doi.org/10.1111/jph.70084 Nguyen, L. L. H., Khuu, D. T., & Nguyen, T. Q. (2023). Factors influencing adoption of precision agriculture. Evaluation Review, 48 (4). https://doi.org/10.1177/0193841X231200775 Odounfa, M. G. F., et al. (2025). Deep learning enables precision agriculture for chili pepper disease detection in Benin. Discover Artificial Intelligence, 5, 315. Omotayo, A. O., Adediran, S. A., Omotoso, A. B., Olagunju, K. O., & Omotayo, O. P. (2025). Artificial intelligence in agriculture: Ethics, impact possibilities, and pathways for policy. Computers and Electronics in Agriculture, 239, 110927. https://doi.org/10.1016/j.compag.2025.110927 Özoğul, G. (2025). Applications of artificial intelligence technologies in agriculture: Advantages, challenges, risks, prospects, and recommendations. Cogent Food & Agriculture, 11 (1). https://doi.org/10.1080/23311932.2025.2568199 Padhiary, M., Saha, D., Kumar, R., Sethi, L. N., & Kumar, A. (2024). Enhancing precision agriculture: A comprehensive review of machine learning and AI vision applications in all-terrain vehicle for farm automation. Smart Agricultural Technology, 8, 100483. https://doi.org/10.1016/j.atech.2024.100483 Parganiha, V., & Verma, M. (2025). An efficient disease prediction in smart agriculture using advanced deep learning methods for improving crop productivity. Journal of Phytopathology. https://doi.org/10.1111/jph.70160 Pearson, S. (2025). Artificial intelligence in horticulture. The Journal of Horticultural Science and Biotechnology, 100 (5), 564–570. https://doi.org/10.1080/14620316.2025.2475401 Rashid, A. B., Kausik, A. K., Khandoker, A., & Siddque, S. N. (2025). Integration of artificial intelligence and IoT with UAVs for precision agriculture. Hybrid Advances, 10, 100458. https://doi.org/10.1016/j.hybadv.2025.100458 Ryan, M., Isakhanyan, G., & Tekinerdogan, B. (2023). An interdisciplinary approach to artificial intelligence in agriculture. NJAS: Impact in Agricultural and Life Sciences, 95 (1). https://doi.org/10.1080/27685241.2023.2168568 Saiz, E., Iqbal, F., Grant, J. H., Korir, L., Ullah, S., & Radu, A. (2026). Technological and socio-economic challenges in the development of sensors for precision agriculture. Agriculture, 16 (1), 89. https://doi.org/10.3390/agriculture16010089 Sood, A., Bhardwaj, A. K., & Sharma, R. K. (2022). Towards sustainable agriculture: Key determinants of adopting artificial intelligence in agriculture. Journal of Decision Systems, 33 (4), 833–877. https://doi.org/10.1080/12460125.2022.2154419 Su, K.-W., Wang, C.-H., Lin, O. Y.-H., Chen, H.-J., Wang, W.-L., & Chiu, P.-C. (2026). Human-centered design of an artificial intelligence-based diagnostic system for tomato diseases in smallholder agriculture. Journal of Industrial and Production Engineering. https://doi.org/10.1080/21681015.2026.2613023 Tran Cao Uy, Limnirankul, B., Kramol, P., Hoang Gia, H., & Nguyen Thi, D. T. (2024). Digital technology adoption among smallholder farmers in Vietnam: Implications for digital agricultural extension strategies. Journal of International Development. https://doi.org/10.1002/jid.3904 Ugwu, O. P.-C., Ogenyi, F. C., Alum, E. U., Eze, V. H. U., Basajja, M., & Ugwu, J. N. (2025). Implementing artificial intelligence and machine learning algorithms for optimized crop management: A systematic review on data-driven approaches to enhancing resource use and agricultural sustainability. Cogent Food & Agriculture, 11 (1). https://doi.org/10.1080/23311932.2025.2569982 Upadhyay, A., Chandel, N. S., Singh, K. P., Chakraborty, S. K., Nandede, B. M., Kumar, M., Subeesh, A., Upendar, K., Salem, A., & Elbeltagi, A. (2025). Deep learning and computer vision in plant disease detection: A comprehensive review of techniques, models, and trends in precision agriculture. Artificial Intelligence Review, 58, 92. https://doi.org/10.1007/s10462-024-11100-x Wu, P., & Zhong, Y. (2025). Artificial intelligence in sustainable agriculture: Towards a socio-technical roadmap. Smart Agricultural Technology, 12, 101578. https://doi.org/10.1016/j.atech.2025.101578 Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Revision Version 1 posted Editorial decision: Revision requested 17 Apr, 2026 Reviews received at journal 26 Mar, 2026 Reviewers agreed at journal 18 Mar, 2026 Reviewers agreed at journal 14 Mar, 2026 Reviews received at journal 11 Mar, 2026 Reviewers agreed at journal 01 Mar, 2026 Reviewers invited by journal 27 Feb, 2026 Editor invited by journal 27 Feb, 2026 Editor assigned by journal 09 Feb, 2026 Submission checks completed at journal 09 Feb, 2026 First submitted to journal 06 Feb, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8808017","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Systematic Review","associatedPublications":[],"authors":[{"id":588364649,"identity":"b6ac6e5d-c0af-4a32-8a9f-b4ebb4453a59","order_by":0,"name":"Sandip Satpati","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA2ElEQVRIiWNgGAWjYHACAzDJxt4AYlsQr0WCjecAiC1BghYGiQQITRDIz27e9vDnDrs6PsnnVzf8KJBg4G/vTsBvxZ1j5ca8Z5Il2KRzym72AB0mcebsBvxaJHLMpBnbmEFa0m7wALUYSOTi1yI/I8dM8mdbvQSb5Jm0m3+I0cJwI8dMgrftsASbBPux20TZYnAjDeSX45JtPDlst2UMJHgI+kV+RjIoxKr55duPP7v55o+NHH97LwGHAeOdgbEBRPOAI4iHkHJkLewPiFE9CkbBKBgFIxAAABcOQgYmU5AKAAAAAElFTkSuQmCC","orcid":"","institution":"Shaheed Bhagat Singh College","correspondingAuthor":true,"prefix":"","firstName":"Sandip","middleName":"","lastName":"Satpati","suffix":""}],"badges":[],"createdAt":"2026-02-06 14:09:37","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8808017/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8808017/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":102746205,"identity":"c10ea4f8-0e3d-48df-901b-4be5de0d09ae","added_by":"auto","created_at":"2026-02-16 08:56:05","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":572639,"visible":true,"origin":"","legend":"\u003cp\u003ePRISMA Diagram\u003c/p\u003e\n\u003cp\u003eSource: By Author\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-8808017/v1/514489a4aeb02a9f3b091f4a.png"},{"id":102491569,"identity":"f37322bb-3d5a-4c8b-b782-31cb58a227a8","added_by":"auto","created_at":"2026-02-12 08:44:49","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":426868,"visible":true,"origin":"","legend":"\u003cp\u003eConceptual Model for Smallholder-Centric AI Adoption Framework\u003c/p\u003e\n\u003cp\u003eSource: By Author\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-8808017/v1/4132895d6e7184e33241591d.png"},{"id":102750581,"identity":"7b5dfa03-da35-42c6-9339-8f34ab99e9f0","added_by":"auto","created_at":"2026-02-16 09:20:36","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2336596,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8808017/v1/360fa700-fb8e-4ec6-9f24-396f1a87e44d.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"AI-Enabled Precision Agriculture for Smallholder Farmers","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003e\u003cstrong\u003e1.1 Background and Rationale\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSmallholder agriculture has continued to be the mainstay of food production in most developing and emerging economies, yet it is increasingly being limited by a complex of structural and environmental issues. The low productivity of smallholder farms has not increased with a lack of access to quality inputs, poor extension services, minor landholdings, and information asymmetry (Aijaz et al., 2025; Mana et al., 2024). The climate change risks such as heightened temperatures, irregular rainfall trends, the increase in pest and disease cases, and soil erosion also compound these limitations and burden the farmers who are resource-constrained (Wu and Zhong, 2025).\u003c/p\u003e\n\u003cp\u003eMoreover, rural labor unavailability caused by urban migration and demographic shifts is exerting a strain on the traditional farm system that is labor-intensive, lowering the efficiency of operations and the ability to manage farms on time (Omotayo et al., 2025). All these problems jeopardize the food security, agricultural incomes, and the sustainability of the agricultural systems based on smallholders, which highlights the necessity of more scalable, adaptive, and data-informed responses.\u003c/p\u003e\n\u003cp\u003ePrecision agriculture (PA) has become one of the most significant paradigms currently that utilize data-driven technologies to empower agricultural decision-making on a fine spatial and temporal level. Recent developments in artificial intelligence (AI), machine learning (ML), computer vision, Internet of Things (IoT), and unmanned aerial vehicles (UAVs), have made considerable changes to the functionality of PA systems, allowing them to monitor in real-time, engage in predictive analytics, and perform automated operations in the field (Bayar et al., 2025; Rashid et al., 2025). The AI-powered PA applications now cover such crucial areas as the detection of crop diseases, yield forecasting, intelligent irrigation, nutrient control, and the evaluation of soil health (Majdalawieh et al., 2025; Mamabolo et al., 2025).\u003c/p\u003e\n\u003cp\u003eCombining the heterogeneous sources of data, such as satellite imagery, proximal sensors, weather data, and farm management records, AI systems can create actionable information to enhance the efficiency of resource use, decrease the input costs, and become more resilient to climatic variability (Aijaz et al., 2025; Padhiary et al., 2024). When considering the aspect of sustainability, AI-based PA can help with environmentally friendly farming by reducing the overuse of fertilizers and pesticides and ensuring or even enhancing productivity (Mana et al., 2024). As a result, AI-enabled PA is becoming one of the foundations of climate-smart and sustainable agriculture.\u003c/p\u003e\n\u003cp\u003eAlthough it has stood to promise great success, the application of precision agriculture has been historically uneven within large-scale commercial operations situated in developed countries where capital accessibility, digital connectivity, and technical know-how are comparatively high (Saiz et al., 2026). Those systems can be based on expensive sensors, custom platforms, sophisticated equipment, and sophisticated data analytics pipelines that can be inaccessible by the majority of smallholder farmers. Subsequently, technological disparity remains, which restricts the inclusiveness and equal influence of PA innovations (Omotayo et al., 2025).\u003c/p\u003e\n\u003cp\u003eAs a contrast, smallholder-oriented AI systems of the future are based on affordability, scalability, and relevance to situations. These systems are moving towards more extensive utilization of low-cost IoT sensors, mobile-based decision support systems, small-scale machine learning models, and cloud-based analytics based on small plots and small-scale farming heterogeneity (Bayar et al., 2025; Wu and Zhong, 2025). In addition, the significance of socio-technical design, usability formulated by farmers, and policy facilitation to guarantee meaningful adoption and long-term effects among smallholder groups has been identified in recent studies (Saiz et al., 2026; Wu and Zhong, 2025). This distinction is important to understand to be able to align AI-enabled precision agriculture with the objectives of inclusive development and make sure that the technology will be converted into the actual benefits that smallholder farmers will receive.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e1.2 Problem Statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAlthough there has been rapid growth and development of artificial intelligence (AI) technologies in agriculture, their uptake among the smallholder farmers is widely uneven and disjointed in terms of regional and production systems. Although AI-based solutions are expected to enhance productivity, sustainability, and resilience, empirical data show that such benefits are over- and under-represented in technologically advanced and well-resourced farming environments (Ahmad et al., 2025; Özoğul, 2025). Still, smallholder farmers are grappling with a lack of digital infrastructure, cost levels of implementation, the lack of technical assistance, and the lack of confidence in algorithmic decision-making, which leads to an ongoing adoption gap between the potential of the technology and its application on the ground (Hiywotu, 2025; Sood et al., 2022).\u003c/p\u003e\n\u003cp\u003eThe major weakness of the current literature is that most studies are based on technology-focused research methods that emphasize approximately accuracy of algorithms, system architecture, and computing capabilities with minimal consideration on socio-economic, behavioral and institutional aspects of adoption. Most of the studies measure AI systems based on technical performance scales, and little of them incorporate the views of a farmer, contextual limitations, or governance factors (Baladraf et al., 2025; Ryan et al., 2023). This imbalance limits the practical applicability of AI innovations because the decisions of adopting such innovations in smallholder agriculture are not only influenced by the technological effectiveness of the technology but also social pressure, usefulness, risk-taking behavior, and conditions of implementation (Lee et al., 2024; Pearson, 2025).\u003c/p\u003e\n\u003cp\u003eFurthermore, the lack of holistic and integrative frameworks, which directly connect the performance of AI systems to the perceptions of farmers and the consequences of subsequent adoptions, is significant. The available research usually investigates these aspects separately, neglecting dynamic relationships amid technological dependability, trust on behalf of the users, and behavioral intention (Ugwu et al., 2025). Participatory and human-centred design methods that have demonstrated the potential to enhance usability and acceptance are not well represented in AI-for-agriculture studies, especially in the smallholder setting (Su et al., 2026). Due to this, the absence of consistent models that can link AI performance, social-economic perception, and adoption behavior constrains theoretical progress as well as policy implications.\u003c/p\u003e\n\u003cp\u003eThe way to bridge these gaps is adhering to interdisciplinary and a socio-technical research paradigm that incorporates the performance assessment of AI with behavioral, economic, and institutional assessments. In the absence of such integrative structures, AI based agricultural solutions can strengthen existing inequalities instead of becoming part of inclusive and sustainable food systems (Ahmad et al., 2025; Ryan et al., 2023). Therefore, the systematic association between AI technological opportunities and farmer-centered adoption processes is urgently required, which would guide scalable, equitable, and context-driven deployment policies in smallholder agriculture.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e1.3 Research Objectives\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eI. To assess the impact of AI tools (e.g., crop disease detection, yield prediction) on agricultural productivity.\u003c/p\u003e\n\u003cp\u003eII. To evaluate how rural farmers perceive and adopt AI-based agricultural technologies.\u003c/p\u003e\n\u003cp\u003eIII. To identify barriers and enablers for implementing AI in smallholder agriculture.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e1.4 Originality and Innovation of the Study.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe current study contributes to the current body of research on artificial intelligence (AI) in agriculture by going beyond the largely technology-focused views and providing a comprehensive, smallholder-focused, and policy-relevant contribution. Although the previous knowledge and practice on the technical potential of AI and machine learning to improve productivity, disease management, and resource efficiency have been well documented (Clay et al., 2024; Gupta et al., 2025; Minhans et al., 2025), the practical applicability of technology efficiency to specific change behavior in the real-world setting of smallholder farmers has been relatively overlooked. This paper fills this gap by explicitly relating the effectiveness of AI tools to human adoption processes, which will further facilitate a more socio-technical interpretation of AI-enabled agriculture.\u003c/p\u003e\n\u003cp\u003eFirst, the research combines AI performance indicators and human-related adoption elements, including trust, perceived usefulness, usability and contextual limits. The literature usually analyzes the AI systems independently, focusing on accuracy, optimization efficiency, or computational progress (Parganiha \u0026amp; Verma, 2025; Adinarayana et al., 2024). Conversely, this study contributes to the growing arguments around the importance of trust-aware and user-friendly AI implementation to agriculture by systematically relating system-level efficiency with agricultural perception and behavioral outcome (Gardezi et al., 2023; Erickson and Fausti, 2021). In such a way, this piece of research will offer both empirical and theoretical data as to why AI solutions that are technically sound can still perform poorly in terms of uptake in the context of smallholders.\u003c/p\u003e\n\u003cp\u003eSecond, the research suggests a Smallholder-Centric AI Adoption Framework, which is one of the theoretical and methodological novelties. In contrast to the generic precision agriculture or smart farming frameworks that are mostly developed based on large-scale and capital-intensive farming systems (Clay et al., 2024; Cardak et al., 2025), the given framework explicitly considers the socio-economic context of the smallholder farmers, such as resource availability, digital literacy, institutional backing, and the extension procedures. Based on the findings of the digital technology adoption research carried out in the context of smallholder, the conceptual framework defines adoption as the dynamic process connecting the performance of AI systems, perception of farmers and adoption outcomes. The contribution is a direct reaction to the calls of interdisciplinary and inclusive AI research in agriculture (Gardezi et al., 2023; Gupta et al., 2025).\u003c/p\u003e\n\u003cp\u003eThird, the research provides policy and design recommendations in practice that go beyond the study of algorithmic innovation. Though a large part of the current body of literature focuses on the technological progress, there are not many studies that convert their results into the recommendations to policymakers, designers, and extension systems (Erickson and Fausti, 2021). This study claims that deficiency by creating design concepts of AI tools suited to smallholder, such as affordability, transparency, and compatibility with extension services, and by explaining policy directions to facilitate inclusive AI diffusion by capacity building, governance systems, and investing in digital infrastructures (Ishore et al., 2025; Tran Cao Uy et al., 2024). In this way, the research helps to fill the gap between the agricultural policy and sustainable development goals and AI research.\u003c/p\u003e\n\u003cp\u003eIn general, this study is novel in its combination of a socio-technical approach, a specific study of smallholder farmers, and an emphasis on transforming AI innovation into inclusive and scalable agricultural transformation. By matching AI performance and adoption dynamics with the relevance of policies, the research advances the existing body of work on precision agriculture and AI-for-agriculture toward more balanced, impactful results.\u003c/p\u003e"},{"header":"2. Precision Agriculture: The Conceptual Underpinnings of AI-Assisted Agriculture.","content":"\u003cp\u003eThe innovation of Precision Agriculture and Smallholder Farming Systems is the second one.\u003c/p\u003e \u003cp\u003ePrecision agriculture (PA): This is the use of digital, data-driven, and sensor-based technologies to optimise agricultural inputs and management decisions on a fine spatial and temporal scale. PA, which started as a part of the large-scale mechanized farming systems, now includes artificial intelligence (AI), Internet of Things (IoT), robotics, and big data analytics to support real-time decisions (Atasoy, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Abdulraheem et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2026\u003c/span\u003e). Although such innovations yield potential efficiency gains and benefits that are sustainable, their creation has been influenced in large measure by the requirements of industrial farming systems.\u003c/p\u003e \u003cp\u003eThe problem of structural differences between the smallholder and industrial farming systems is a fundamental factor affecting the applicability of PA technologies. Industrial farms are usually located on large, uninterrupted areas of land, highly mechanized and capitalized and equipped with digital infrastructures, which allow a smooth connection of sophisticated PA instruments (Almazmomi, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Conversely, smallholder systems are typified by fragmented plots, diversified production activities, low levels of mechanization, and use of family labor that makes it hard to direct transfer conventional PA models (Nguyen et al., \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Lankamo et al., \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThere are also limitations that restrict the PA adoption by the smallholders. Poor access to capital and credit will limit investment in sensors, equipment and connectivity, and poor digital infrastructure and network coverage will also complicate the implementation of technology (Sharma et al., 2025). The lack of skills and digital illiteracy limits access to information and access to and trust in data-driven recommendations, which highlights the role played by extension and capacity-building mechanisms (Daum, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; High et al., \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Moreover, digital instruments are not usually created to fit the context of a smallholder and are therefore less usable and relevant (Dittmer et al., \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). These limitations lead to the necessity of adaptive, inclusive, and smallholder-oriented PA frameworks.\u003c/p\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e2.2 Artificial Intelligence in Precision Agriculture\u003c/h2\u003e \u003cp\u003eModern precision agriculture has established artificial intelligence (AI) as a core element to optimize crop management and resource utilization based on data. Both machine learning (ML) and deep learning (DL) are utilized all over the yield prediction, soil property estimation, climate risk modeling, and input optimization by learning complex, non-linear relationships using multi-source data in agriculture (Aijaz et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Ugwu et al., \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Recent developments of deep neural networks have also increased prediction capability and scalability, application in adaptive decision-making in changing field conditions (Mana et al., \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Wu and Zhong, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eOne of the most developed uses of AI in precision agriculture is computer vision, in crop disease and pest detection. Vision-based classifiers and convolutional neural networks allow the timely detection of diseases in the form of images of leaves and field imagery and shorten the time of diagnosis, and decrease the losses of yields (Majdalawieh et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Pearson, \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). These strategies are being used more and more on mobile gadgets and edge computers and enhance their usability in the real-time field (Padhiary et al., \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Su et al., \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2026\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe combination of AI and the Internet of Things (AIoT), unmanned aerial vehicles (UAVs), proximal sensors, and new 5G connectivity have greatly expanded the space and time resolution of farm surveillance. AIoT can be used to facilitate smart irrigation, nutrient control, and stress detection using sustained data flows of distributed sensors and aerial stations (Bayar et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Rashid et al., \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). The affordability of sensors, interoperability, and connectivity are, however, major challenges, especially in the smallholder setting (Saiz et al., \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2026\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThese technologies meet at AI-based decision support systems and advisory systems that turn complex analytics into actionable recommendations. These systems are vital in the process of bridging the technology proficiency with farmer choice, focusing on usability, confidence, and socio-technical integration (Lee et al., \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Ahmad et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Ryan et al., \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003e2.3 Theoretical Lenses for Technology Adoption\u003c/h2\u003e \u003cp\u003eAdoption of technologies based on artificial intelligence (AI) into agriculture can be viewed in a number of complementary theoretical perspectives that reflect the individual, social and systemic aspects of technological change. The Technology Acceptance Model (TAM) is an initial system that offers a basis with the application of the perceived usefulness and perceived ease of use as key factors that determine technology adoption. Empirical research in the field of agricultural digitalization has shown that the intention to use precision and AI-based applications is largely affected by the expectation of its productivity, income growth, and the decreased amount of labor and risks by farmers (Nguyen et al., \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Sharma et al., 2025). Nevertheless, perceived ease of use is still limited by digital skills and access, especially in the case of smallholder farmers (Daum, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Dittmer et al., \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eIn addition to the perceptions of an individual, Diffusion of Innovations theory places emphasis on social learning, communication medium, and institutionalization in influencing the adoption pathways. The impact of agricultural extension services, peer networks, and demonstration effects is substantial regarding the issue of the spread of innovations in the world of farmers (Lee et al., \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Lankamo et al., \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Research on AI-assisted extension systems also suggests that the process of diffusion is hastened when technologies are consistent with local practices and promote transformative learning and do not involve mere transfer of information (High et al., \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe newer research focuses on human-centered AI and socio-technical systems, acknowledging that the use of technology is within a wider social, cultural, and organizational framework. Human-centered design models are focused on usability, inclusivity, and contextual relevance, which means that AI systems assist farmers in decision-making, but do not take the decision-making powers away (Ryan et al., \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Su et al., \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2026\u003c/span\u003e). Socio-technical approaches also emphasize the interrelationship between technological systems, governing relationships, and user capacities, especially in agricultural regimes characterised by resource-based constraints (Atasoy, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Baladraf et al., \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eLastly, the concepts of trust, explainability, and ethics have become important factors in AI adoption in agriculture. Issues of data ownership, algorithmic secrecy, environmental sustainability as well as social fairness may erode user confidence unless properly resolved (Ahmad et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; \u0026Ouml;zoğul, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Responsible and inclusive adoption of explainable and transparent AI systems with the help of ethical governance frameworks and secure data architectures is thus important (Almazmomi, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Abdulraheem et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2026\u003c/span\u003e; Neethirirajan, 2025). These theoretical perspectives have a synthesis that offers a wholesome framework of analyzing AI uptake in precision agriculture in terms of technical, social, and ethical aspects.\u003c/p\u003e \u003c/div\u003e"},{"header":"3. Methodology","content":"\u003cp\u003e\u003cstrong\u003e3.1 Data Sources and Selection Criteria.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe review had a systematic and clear literature selection process that guaranteed the inclusion of all scholarly literature on artificial intelligence (AI)-enabled precision agriculture with the particular aim of the smallholder farming system and technology adoption. Five large academic publishers and databases with a reputation of high-quality agricultural and interdisciplinary research were searched: Elsevier/ScienceDirect, Taylor and Francis, Wiley Online Library, SAGE Publications and Springer Nature retrieved peer-reviewed journal articles and book chapters. The rationale behind these platforms lies in the fact that they are highly represented in journals listed in the Scopus database and cover the issues of AI, precision agriculture, and rural development research comprehensively (Clay et al., 2024; Ahmad et al., 2025).\u003c/p\u003e\n\u003cp\u003eThe period of the review was 20172026, which reflects the dynamism of machine learning, deep learning, AIoT, and socio-technical adoption research in agriculture, especially after the advent of Agriculture 4.0 and 5.0 paradigms (Sood et al., 2022; Baladraf et al., 2025). (Artificial intelligence, precision agriculture, machine learning, smallholder farmers, technology adoption, decision support systems, and AIoT) were some of the keywords that were used to search the material.\u003c/p\u003e\n\u003cp\u003eInclusion criteria: The studies needed to fulfill the following inclusion criteria:\u003c/p\u003e\n\u003cp\u003e(i) AI-based technologies (e.g., machine learning, deep learning, computer vision, AIoT) were explicitly applied and reviewed;\u003c/p\u003e\n\u003cp\u003e(ii) were located in the precision or smart agriculture contexts;\u003c/p\u003e\n\u003cp\u003e(iii) dealt with adoption, usability, social-economic effects or smallholder agricultural regimes empirically or conceptually. Articles that did not involve the development of the algorithm on the laboratory level and were not followed up with the applications in agriculture, non-peer-reviewed materials, and the articles that did not involve the adoption or farm-level decision-making were excluded.\u003c/p\u003e\n\u003cp\u003eThe last corpus contained systematic reviews, empirical researches, and theoretical frameworks that cover AI implementation, governance, ethics, and adoption processes in different agro-ecological and socio-economic settings (Aijaz et al., 2025; Nguyen et al., 2023; Lankamo et al., 2025; Pacal et al., 2024). This strategy provided analytical rigor and also stayed relevant to smallholder-focused precision agricultural research.\u003c/p\u003e\n\u003cp\u003eTable 1: Classification of Reviewed Articles by Major Themes\u003c/p\u003e\n\u003ctable border=\"1\" cellpadding=\"0\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eTheme\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eArticles (Author, Year)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eAI Application / Focus\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eDimension Addressed\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eKey Findings\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e1. AI \u0026amp; Technology in Crop Management\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eAijaz et al., 2025; Bayar et al., 2025; Mamabolo et al., 2025; Majdalawieh et al., 2025; Padhiary et al., 2024; Rashid et al., 2025; Upadhyay et al., 2025; Sudha \u0026amp; Loret, 2026; Odounfa et al., 2025; Filippi et al., 2025; Mahale et al., 2024; Kuradusenge et al., 2024; Ajith et al., 2025; Gupta \u0026amp; Kumar Pal, 2025; Minhans et al., 2025; Parganiha \u0026amp; Verma, 2025; Clay et al., 2024; Gupta et al., 2025\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003ePrecision agriculture, ML/DL, UAVs, IoT, sensors, yield prediction, disease detection\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTechnology / System\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eAI improves crop productivity, accuracy of disease detection, and yield prediction across smallholder and commercial contexts\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e2. Human / Adoption \u0026amp; Smallholder-Centric Factors\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eOmotayo et al., 2025; Wu \u0026amp; Zhong, 2025; Ahmad et al., 2025; Lee et al., 2024; Sood et al., 2022; Su et al., 2026; Tran Cao Uy et al., 2024; Dittmer et al., 2025; High et al., 2025; Nguyen et al., 2023; Sharma et al., 2025; Lankamo et al., 2025\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eAdoption determinants, participatory design, extension, smallholder engagement\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eHuman / System\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eHuman factors such as trust, skills, participatory design, and access to digital services significantly influence AI adoption\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e3. System, Policy \u0026amp; Governance\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eMana et al., 2024; Saiz et al., 2026; Abdulraheem et al., 2026; Almazmomi, 2025; Atasoy, 2025; Omotayo et al., 2025; Erickson \u0026amp; Fausti, 2021\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003ePolicy frameworks, socio-economic challenges, infrastructure, food security\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eSystem / Technology\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eAI adoption depends on supportive policy, digital infrastructure, and socio-economic context; governance frameworks are essential\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e4. Sustainability \u0026amp; Socio-Technical Integration\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eBaladraf et al., 2025; Hiywotu, 2025; Ryan et al., 2023; Ahmad et al., 2025; Wu \u0026amp; Zhong, 2025; Mana et al., 2024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eAgriculture 4.0/5.0, sustainable agriculture, socio-technical integration\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTechnology / Human / System\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eIntegration of AI with human and system factors promotes sustainable agriculture and enhances adoption in developing contexts\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eSource: By Author\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.2 Review Protocol\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA PRISMA-based (Preferred Reporting Items of Systematic Reviews and Meta-Analyses) screening protocol was used in this review to guarantee the methodological transparency, reproducibility, and rigor in the selection and synthesis of the relevant literature (Figure 1). A preliminary list of records was obtained in Elsevier/ScienceDirect, Taylor \u0026amp; Francis, Wiley, SAGE and Springer Nature databases. Titles and abstracts were filtered after the elimination of duplicates and used to remove irrelevant material according to the criteria of being relevant to AI-enabled precision agriculture, and especially focusing on material relating to smallholder farming systems and technology adoption. The next step implied full-text screening, which ensured that the obtained corpus of peer-reviewed articles complies with the established inclusion criteria, thus a narrowed down set of publications that can be included in qualitative synthesis (Majdalawieh et al., 2025; Pacal et al., 2024).\u003c/p\u003e\n\u003cp\u003eAfter the screening, thematic coding and synthesis approach was used. Articles were coded inductively and deductively to reflect on recurring conceptual, technical, and socio-institutional concept(s) of AI deployment in precision agriculture. The categories were repeatedly revised to include technological dimensions (e.g., algorithm performance, sensing accuracy) and human-centered dimensions (e.g., adoption drivers, trust, governance), which aligns with socio-technical views highlighted in recent AI-agriculture research (Wu \u0026amp; Zhong, 2025; Ryan et al., 2023).\u003c/p\u003e\n\u003cp\u003eAccording to thematic convergence, the reviewed articles were categorized as three analytical groups. The former included technical performance studies, which concerned machine learning, deep learning, computer vision, AIoT, UAVs, and sensor-based yield prediction, disease detection, and resource optimization systems (Aijaz et al., 2025; Rashid et al., 2025; Upadhyay et al., 2025).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe second group consisted of adoption and perception studies that investigated farmer attitudes and behavior intentions, institutional support, and socio-economic factors that impact the adoption of AI-based precision agriculture, especially in the case of smallholders (Nguyen et al., 2023; Tran Cao Uy et al., 2024; Lankamo et al., 2025). The third group included the study of policy, ethics and sustainability with a focus on governance systems, ethical risks, explainability, equity and long-term sustainability of AI-driven agricultural systems (Omotayo et al., 2025; Ahmad et al., 2025; Gardezi et al., 2023).\u003c/p\u003e\n\u003cp\u003eThis systematic procedure facilitated a combined synthesis of technical performance, human adoption procedures, and policy-related findings that is consistent with fresh demands of comprehensive and conscientious AI frameworks in precision agriculture.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.3 Analytical Framework\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe analytical model used to develop this research is implemented in the form of a three-layered model, whereby technological, human, and institutional areas are incorporated to determine the adoption and the effect of AI-enabled precision agriculture among smallholder farmers. This framework enables one to understand in depth how AI technologies affect productivity, farmer behavior, and the enabling environment needed to make the adoption sustainable.\u003c/p\u003e\n\u003cp\u003eI. Artificial Intelligence and Productivity Reports.\u003c/p\u003e\n\u003cp\u003eThe initial layer dwells upon AI tools and their direct contribution to agricultural productivity. These are machine learning algorithms, computer vision, Unmanned Aerial Vehicles (UAVs), AIoT systems, and sensor networks that detect crop diseases, predict yields, monitor soil, optimize irrigation, and manage pests (Majdalawieh et al., 2025; Bayar et al., 2025; Rashid et al., 2025). Empirical and review literature proves that these tools are more efficient in terms of resources, have less input waste, and provide more efficient interventions to crop health and productivity (Aijaz et al., 2025; Mamabolo et al., 2025; Padhiary et al., 2024). In addition, a combination of AI and IoT and UAV can also be used to enable real-time monitoring and automated decision-making, which can play a role in resilient and sustainable production systems (Mana et al., 2024; Wu and Zhong, 2025).\u003c/p\u003e\n\u003cp\u003eII. Farmer Perception and Adoption Behavior\u003c/p\u003e\n\u003cp\u003eThe second level looks into the perceptions, attitudes and adoption behavior of farmers. The perceived usefulness, ease of use, social influence, and economic incentives are essential factors to successful technology uptake, in addition to the technical performance (Lee et al., 2024; Sood et al., 2022; Tran Cao Uy et al., 2024). The awareness, digital literacy, risk perception, and access to the advisory services are the factors that influence the intention of the smallholder farmers to adopt AI-based tools (Nguyen et al., 2023; Su et al., 2026; High et al., 2025). The knowledge of these behavioral determinants is essential in developing human-centered AI systems that will meet the needs, capacities, and local farming practices of farmers (Ryan et al., 2023; Pearson, 2025).\u003c/p\u003e\n\u003cp\u003eIII. Policy, Infrastructure and Institutions Enabling Environment.\u003c/p\u003e\n\u003cp\u003eThe third layer underlines the enabling environment that facilitates the use of AI in agriculture. This involves the national and regional policies, the digital infrastructure, the institutional capacity, the extension services, and the ethical, transparent, and sustainable AI deployment frameworks (Ahmad et al., 2025; Omotayo et al., 2025; Gardezi et al., 2023). Research points out that regulatory directions, ICT infrastructure investment, and fair access to the technology are essential to implement AI interventions that are inclusive and scalable to smallholders (Daum, 2025; Dittmer et al., 2025; Atasoy, 2025). Also, social-technical integration will promote the fact that AI applications are not merely technologically efficient but also socially and economically feasible in local farming settings (Gupta and Kumar Pal, 2025; Ugwu et al., 2025).\u003c/p\u003e\n\u003cp\u003eIntegrated Perspective\u003c/p\u003e\n\u003cp\u003eA combination of these three layers grants the framework a multifaceted perspective to evaluate AI-enabled precision agriculture. It also connects technical performance with adoption conduct and aligns both of them in a conducive institutional setting. In this way, the leverage areas of enhancing productivity, raising adoption rates among smallholders, and advancing sustainable and ethical AI use in the agricultural sector can be identified.\u003c/p\u003e"},{"header":"4. Impact of AI Tools on Agricultural Productivity","content":"\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003e4.1 AI for Crop Disease Detection\u003c/h2\u003e \u003cp\u003eCNNs, Vision and UAV-based Monitoring.\u003c/p\u003e \u003cp\u003eThe latest achievements in the field of artificial intelligence (AI) have made convolutional neural networks (CNNs) and methods of computer vision the fundamental drivers of automated crop disease detection. ResNet, DenseNet, EfficientNet, and customized lightweight architectures are also deployed on the leaf-level and canopy-level classification of disease based on CNN-based architectures because they are highly effective in feature extraction and generalization (Majdalawieh et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Upadhyay et al., \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Pacal et al., 2024). RGB, multi spectrum, and hyperspectral based vision systems have been shown to be highly diagnostic when used together with transfer learning and data augmentation methods under controlled and semi-controlled settings (Aijaz et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Minhans et al., \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe Unmanned Aerial Vehicles (UAVs) also expand AI-based disease detection capabilities beyond plot-scale surveillance to the field-scale surveillance allowing high-resolution data collection in real-time over large agricultural plots. AI systems using UAVs in combination with CNNs and Internet of Things (IoT) and Artificial Intelligence of Things (AIoT) platforms facilitate the prevention of diseases before they occur, spatial disease mapping, and the use of targeted intervention plans (Bayar et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Rashid et al., \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Such systems are becoming more consistent with the precision agriculture paradigms, and dynamic disease-monitoring is possible, with less reliance on labor and the use of chemicals (Gupta and Kumar Pal, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Sudha and Loret, 2026).\u003c/p\u003e \u003cp\u003ePerformance Measures vs. actual scalability.\u003c/p\u003e \u003cp\u003eAlthough CNN-based disease detection models may claim high performance, including accuracy, precision, recall, F1-score, and area under the curve (AUC), their applicability to large-scale learning is limited by the heterogeneity of data sources and their environmental variability and scalability (Majdalawieh et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Filippi et al., \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Most models are trained using curated data with homogeneous lighting, backgrounds and growth of crops, thus restricting their robustness when deployed to a wide field of application (Pacal et al., 2024; Wu and Zhong, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eComputational issues, energy usage, connection density, and rural access to edge or cloud systems also aggravate the problem of scalability (Saiz et al., \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2026\u003c/span\u003e; Omotayo et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Despite the promising results of lightweight CNNs and edge-AI solutions, trade-offs between the complexity of models and diagnostic accuracy still play a major role (Padhiary et al., \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Parganiha and Verma, \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Therefore, recent articles state that explainable, energy-efficient, and context-aware AI models with a focus on balancing the predictive performance and operational feasibility in actual farming settings are needed (Gardezi et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; \u0026Ouml;zoğul, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eAsian and African Case Studies.\u003c/p\u003e \u003cp\u003eAsian and African experience shows the promise and the constraints of AI-based systems of crop disease detection in smallholder-managed agricultural systems. CNN-based vision systems have been put into use to date with crops in South and Southeast Asia that include rice, tomato, wheat, and chili with significant improvements in the accuracy of identifying diseases and supporting farmers in decision-making (Ajith et al., \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Su et al., \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2026\u003c/span\u003e; Mahale et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Nevertheless, digital literacy, the availability of extension services, and socio-economic preparedness are the powerful factors influencing the adoption (Tran Cao Uy et al., \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Sharma et al., 2025).\u003c/p\u003e \u003cp\u003eUAV- and smartphone-based AI-based disease detection systems have been implemented in African contexts to deal with labour shortages and lack of agronomic knowledge especially with staple and horticultural crops. Benin and Ethiopian case studies demonstrate that deep learning models are effective to recognize fungal and viral diseases in the actual field conditions, which can help to protect crops and stabilize yield (Odounfa et al., \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Lankamo et al., \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). However, there are still obstacles, such as data scarcity, affordability, trust, and institutional backing, that do not allow large-scale use (Mamabolo et al., \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Daum, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eAltogether, Asian and African studies highlight the idea that technological performance is not enough, effective implementation of AI-based crop disease detection has to be consistent with local agronomic solutions, inclusive design, and conducive policy and governance frameworks (High et al., \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Ahmad et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Wu and Zhong, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003e4.2 AI-Based Yield Prediction and Resource Optimization\u003c/h2\u003e \u003cp\u003eML/DL Models for Yield Forecasting\u003c/p\u003e \u003cp\u003eThe use of machine learning (ML) and deep learning (DL) has emerged as the focal point of crop yield prediction as they are able to cause complex, nonlinear interactions between climatic, soil, crop, and management variables. Conventional ML methods include random forests, support vector machines, gradient boosting, and k-nearest neighbors, which have been extensively used to predict the yields using historical yield data, meteorological, and soil attributes (Aijaz et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Gupta and Kumar Pal, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). More recently, the predictive performance of DLs, such as artificial neural networks (ANNs), convolutional neural networks (CNNs), and recurrent neural networks (RNNs), has proven to be the best when working with high-dimensional data and/or spatiotemporal data (Ajith et al., \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Filippi et al., \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eLSTM networks are becoming more popular in yield forecasting because they can recreate time-dependent correlations in weather and crop growth cycles and enhance the accuracy of the forecast in a variety of seasons (Mahale et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Kuradusenge et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). The hybrid systems that involve satellite images, UAVs and IoT-based sensors, and ML/DL models augment the spatial resolution and real-time yield estimation further, as yield predictions will correspond to the goals of precision agriculture (Rashid et al., \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Sudha \u0026amp; Loret, 2026). Regardless of these developments, issues concerning the quality and transferability of data across agroecological areas and the interpretability of models of these models remain, which restrict a large-scale operational implementation (Filippi et al., \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Wu and Zhong, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eAgricultural and Nutrient Management Exactness.\u003c/p\u003e \u003cp\u003eAI-assisted resource optimization is crucial to precision irrigation and nutrient management because the technology allows making site-specific and demand-driven decisions. Decision support systems based on MLs combine soil moisture sensors, weather forecasts, crop growth models, and estimations of evapotranspiration to automatically schedule irrigation to minimize water wastage and preserve or improve crop yields (Bayar et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Adinarayana et al., 2024). Fuzzy logic methods and reinforcement learning have also been investigated in the field of adaptive control of irrigation in changing climatic conditions (Mana et al., \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Sudha and Loret, 2026).\u003c/p\u003e \u003cp\u003eOn the same note, AI-based nutrient management systems use ML algorithms to forecast nutrient deficiencies and suggest the best rates of fertilizer application depending on the soil characteristics, crop development, and yielding objectives (Mamabolo et al., \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Ajith et al., \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). AIoT-based systems allow the real-time operation of nutrient delivery and close-loop control to improve nutrient utilization and reduce negative externalities of the environment in terms of leaching and greenhouse gases (Bayar et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Gupta et al., 2025). Nevertheless, sensors are not completely reliable, cost calibration and poor digital infrastructure are major setbacks to the large scale adoption, especially in the small holder farming systems (Saiz et al., \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2026\u003c/span\u003e; Daum, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eData on Gains in Productivity and Efficiency of Input.\u003c/p\u003e \u003cp\u003eAccording to the empirical observations of different agroecological settings, AI-based yield forecasting and improvement of resource optimization can result in quantifiable productivity increase and input efficiency enhancement. Research provides increasing yields due to decisions based on AI support in moderate to large increments, depending on the type of crops, availability of data, and size of implementation (Aijaz et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Erickson and Fausti, 2021). ML-powered precision irrigation systems have continued to be shown as reducing water consumption without affecting yields, which leads to better water productivity and climate resilience (Bayar et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Mana et al., \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2024\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eNutrient optimization made with the assistance of AI has also been linked to reduced fertilizer use and improved use of nutrients, which supports the goals of economic and environmental sustainability (Mamabolo et al., \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Ugwu et al., \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). However, the sizes of achieved benefits greatly depend on the socio-economic factors, institutional backing, and the ability of farmers, to read and believe AI advice (Gardezi et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Lee et al., \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). To ensure a long-term productivity increase, the technological innovation needs to be accompanied by an inclusionary governance system, extension services, and moral deployment models (as highlighted in recent policy-oriented studies) (Omotayo et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Ahmad et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Wu and Zhong, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003e4.3 Sustainability and Environmental Outcomes\u003c/h2\u003e \u003cp\u003eReduced Chemical Use\u003c/p\u003e \u003cp\u003ePrecision agriculture grounded in the use of artificial intelligence has been shown to have a lot of potential as far as the diminishing usage of chemical inputs like pesticides, herbicides, and synthetic fertilizers using the data-driven and precise interventions. The early detection of pest infestations, nutrient deficiencies, and crop stress allows the use of agrochemicals site-specifically and in time by AI-based crop monitoring systems based on machine learning, computer vision, and sensor networks (Majdalawieh et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Padhiary et al., \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Providing the AI-based solution of decision support systems can reduce unnecessary chemical application and still ensure crop productivity by replacing blanket field-level application with variable-rate and spot-specific applications (Aijaz et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Mamabolo et al., \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eDisease detection based on the vision and AIoT platforms also help to reduce chemicals by allowing preventive and not reactive measures in crop protection (Bayar et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Rashid et al., \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). According to the empirical reviews, these strategies help to reduce the risks of environmental contamination and decrease the number of chemicals entering the adjacent ecosystems (Mana et al., \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Gupta and Kumar Pal, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). However, AI-guided reduction plans are successful depending on the sensor precision, algorithm and farmer confidence in the AI advice, which is not evenly distributed in the regions (Saiz et al., \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2026\u003c/span\u003e; Gardezi et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eWater efficiency and Soil health.\u003c/p\u003e \u003cp\u003eThe technologies of AI-based precision agriculture have a direct effect on the enhanced water-use efficiency and better soil health through the optimization of irrigation timings, fertilizer application, and soil management methods. Soil moisture sensors, weather data sensors, and crop growth indicators together in machine learning models make it possible to make irrigation decisions based on real crop demand to avoid over-irrigation and soil degradation (Bayar et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Adinarayana et al., 2024). These systems contribute to managing water sustainably because they reduce the amount of evapotranspiration losses and counteract the threat of salinization of intensive lands (Aijaz et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Sudha \u0026amp; Loret, 2026).\u003c/p\u003e \u003cp\u003eOn the same note, AI-based systems to monitor soil health will support real-time evaluation of soil properties, allowing to balance nutrient levels in the soil and enhance soil retention of organic matter (Mamabolo et al., \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Ajith et al., \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). AI applications can save the soil and prevent soil erosion by encouraging the efficient use of inputs, which helps to maintain the long-term soil productivity (Wu and Zhong, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Nevertheless, the scalability of these advantages can be limited in smallholder-based systems because of socio-economic factors, such as access to digital infrastructure, high costs of sensor deployment, and similar aspects (Daum, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Saiz et al., \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2026\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eClimate-Smart Agriculture\u003c/p\u003e \u003cp\u003eIn response to the concept of climate-smart agriculture (CSA), AI-enabled agricultural systems are increasingly compatible with the principles of climate-smart agriculture (CSA) through boosting productivity, helping farmers to adapt to climate variability, and decreasing the emission of greenhouse gases. Sound AI climate resilience AI employs historical climatic data and real-time environmental provisions to make adaptive decisions in farm management in response to uncertain climatic conditions (Wu and Zhong, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Vijayakumar et al., 2025). Accurate input control through AI helps to meet goals of mitigation by decreasing the energy-intensive consumption of fertilizers and decreasing emissions of wasteful irrigation and the use of chemicals (Mana et al., \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Omotayo et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eAdditionally, the monitoring and predictive systems supported by AI contribute to CSA because they allow evidence-based design of policies and climate-related extension services (Hiywotu, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; High et al., \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Even within this alignment, scholars underline that the environmental sustainability of AI in agriculture is subject to inclusive governance systems, ethical applications and the incorporation of local knowledge to prevent the establishment of inequalities and technological lock, in (Ahmad et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Atasoy, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). This means that the potential of AI in achieving complete climate-smartness cannot be achieved alone through technological, institutional, and socio-economic interventions (Wu and Zhong, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Lankamo et al., \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e"},{"header":"5. Farmer Perception and Adoption of AI Technologies","content":"\u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003e5.1 Awareness, Attitudes, and Behavioral Intentions\u003c/h2\u003e \u003cp\u003eAwareness, Attitudes, and Behavioral Intentions\u003c/p\u003e \u003cp\u003eDigital literacy and AI Trust.\u003c/p\u003e \u003cp\u003eDigital literacy is an important factor when determining awareness and trust of farmers in agricultural technologies based on artificial intelligence (AI). Few digital skills can make farmers unable to process AI-generated knowledge, which will result in a lack of trust in algorithm-based recommendations in fields like disease diagnosis, irrigation planning, and nutrient management (Daum, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Aijaz et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). On the other hand, the increased digital competence allows farmers to be more certain about their data-driven decision-making and to use AI-enhanced tools of precision agriculture more knowledgeably (Wu and Zhong, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Gupta and Kumar Pal, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eTransparency, data ownership and privacy are also factors that determine trust in AI systems. To gain trust in AI applications, the farmers are likely to be more convinced that systems are consistent with agronomic knowledge and give explanable recommendations instead of black box predictions (Gardezi et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Omotayo et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Research highlights the importance of socio-technical integration as a way to boost trust and promote the intention to adopt AI (when AI supplements the experience of farmers), as this concept can strengthen trust and foster positive intentions (Ryan et al., \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Ahmad et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). The participatory design and training interventions play a crucial role especially in the smallholder settings where trust and fear of AI technologies are likely to be built (High et al., \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Su et al., \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2026\u003c/span\u003e).\u003c/p\u003e \u003cp\u003ePerceived usefulness and ease of use.\u003c/p\u003e \u003cp\u003eAttitudes of farmers to the adoption of AI technologies are largely shaped by the perceived usefulness. It is empirically proven that farmers tend to implement AI-based solutions in situations where tangible benefits such as better crop productivity, enhanced input utilization, early disease diagnosis, and minimized production risks are evident (Aijaz et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Ugwu et al., \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Use of AI-based crop protection solutions, intelligent irrigation, and yield prediction have been demonstrated to be effective in improving the operational efficiency and accuracy of decisions, supporting positive attitude towards adoption (Bayar et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Majdalawieh et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eAdoption intentions are also further moderated by ease of use mostly amongst farmers who have limited exposure to technology. The absence of contextual customization, demanding high levels of learning, and complex interfaces can demoralize adoptions despite perceived high usefulness (\u0026Ouml;zoğul, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Saiz et al., \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2026\u003c/span\u003e). Customized systems design, understandable user interfaces, and localized implementation of deployment plans would dramatically enhance the usability and adoption of AI technologies (Su et al., \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2026\u003c/span\u003e; Sudha and Loret, 2026). These conclusions can be correlated with the views offered by the technology acceptance, where usefulness and ease of use have a combined effect on the behavioral intentions of farmers regarding AI-enabled precision agriculture (Sood et al., \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Lee et al., \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2024\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eExtension services and peer networks.\u003c/p\u003e \u003cp\u003eAgricultural extension services can be a key factor in determining the attitude and behavior awareness of farmers in terms of using AI technologies. The extension agents serve as intermediaries that convert the complex AI outputs into actionable and farm-level recommendations to lower cognitive and informational barriers to adoption (Lee et al., \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; High et al., \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). There are indicators that extension-based demonstrations, capacity-building practices, and field trial can have great impacts at raising farmers awareness and confidence in AI applications (Hiywotu, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Vijayakumar et al., 2025).\u003c/p\u003e \u003cp\u003ePeer networks also have the effect of social learning and diffusion of trust as a means of influence in adoption. The farmers are more likely to use AI technologies once they have seen positive results in other neighbors or even known people in the community (Nguyen et al., \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Tran Cao Uy et al., \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Positive attitudes and AI normalization among local agricultural systems are supported through informal peer interactions, cooperatives, and digitally integrated groups of farmers (Wu and Zhong, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Lankamo et al., \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). This interaction between the extension services and peer networks therefore forms an enabling ecosystem that facilitates the continuous and inclusive integration of AI technologies in agriculture.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003e5.2 Socio-Economic Determinants of Adoption\u003c/h2\u003e \u003cp\u003eAgricultural Area, Agricultural Revenue and Education.\u003c/p\u003e \u003cp\u003eThe size of farms is one of the most repeatedly found factors affecting the use of AI and precision agriculture technologies. Bigger farms are also more likely to take on AI-based solutions sooner because of their higher financial capabilities, economies of scale, and capacity to take risks related to high initial costs and uncertain payoff (Aijaz et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Mana et al., \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). According to the studies published in Smart Agricultural Technology and Journal of Agriculture and Food Research, medium- and large-scale commercial farms are more likely to use AI-enabled irrigation systems, disease detection systems, and decision-support tools than smallholders (Bayar et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Mamabolo et al., \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eFarm size is closely interacting with income level, determining the access to AI infrastructure and the long-term sustainability of its use. Farmers with higher incomes have a chance to invest in sensors, AIoT systems, UAVs, and analytics services based on a subscription (Rashid et al., \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Omotayo et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). In the low-income situation, on the other hand, smallholder farmers tend to be credit constrained, have less access to digital infrastructure, and high opportunity costs, which slack adoption despite any gain in productivity (Saiz et al., \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2026\u003c/span\u003e; Sharma et al., 2025).\u003c/p\u003e \u003cp\u003eDigital skills and education come in as a huge mediator of how the availability of technology is related to actual use. It is empirically tested and reviewed that farmers having a higher level of formal education and having been introduced to digital tools earlier have a stronger perceptual attitude towards usefulness and ease of use, which increases faster AI adoption decisions (Daum, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Nguyen et al., \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). In Global South, there is still low digital literacy as a structural obstacle, especially in cases where AI systems use sophisticated interfaces or ability to process data (Wu and Zhong, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Tran Cao Uy et al., \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2024\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eGender and Generation Dimensions.\u003c/p\u003e \u003cp\u003eThe issue of gender inequality is vital in the process of influencing the patterns of adoption of AI in the agricultural sector. Women farmers, particularly in developing areas are frequently unequally supplied with land, credit, training, and digital devices, limiting their involvement with AI-facilitated agriculture (Dittmer et al., \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Lankamo et al., \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Research highlights that women are restricted in their access to digital extension programs and technology tests by socio-cultural norms and institutes even in instances when AI tools are technically available (High et al., \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Ahmad et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe differences between generations also affect the adoption. Farmers of younger age are more willing to use AI, data-driven decision-making, and automation, which is explained by being more digitally familiar and risk-takers (Sood et al., \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Gardezi et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Conversely, more senior farmers tend to be guided by experience and can view AI systems as intricate or incompatible with the existing ways of doing things, which lowers the intention to adopt (Ozoglu, 2025). Nonetheless, it is believed that intergenerational learning and peer influence may help to reduce them, at least, when AI tools are created in the framework of human-centered and participatory design (Su et al., \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2026\u003c/span\u003e; Lee et al., \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2024\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eRegional Inequality and the Global South\u003c/p\u003e \u003cp\u003eRegional inequalities are still acute in the spread of AI technologies, as their usage is not only concentrated in countries with high income but also in developed agricultural areas. Structural issues, including poor digital connectivity, poor signal connectivity, high sensor prices, and divided land parcels are among the major barriers to adoption in the Global South (Vijayakumar et al., 2025; Abdulraheem et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2026\u003c/span\u003e). African and South Asian studies indicate that although AI applications are no longer seen as a limited pilot project with questionable possibilities, many of them are still in the pilot phase (Mamabolo et al., \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Odounfa et al., \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe nature of AI technologies adapted is also determined by socio-economic inequalities across regions. Mobile-based advisory systems and low-cost decision-support tools are more likely to receive adoption in low-and-middle-income countries compared to capital-intensive automation and robotics (Nguyen et al., \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; High et al., \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Moreover, the policy gaps, ineffective extension, and restrained institutional assistance contribute to the regional disparities, suggesting the necessity of the inclusive AI governance and region-specific deployment strategies (Omotayo et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Atasoy, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eOn the whole, the literature tends to agree that AI implementation in agriculture is not a technological process but a profoundly socio-economic phenomenon, which is determined by the characteristics of farms, social systems, and developmental trends. These determinants will have to be addressed when it comes to providing equitable and sustainable AI-driven agricultural transformation, especially in the Global South (Wu and Zhong, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Ryan et al., \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003e5.3 Human-Centered AI and User Experience\u003c/h2\u003e \u003cp\u003eHuman-centered artificial intelligence (HCAI) has been proposed as a more important paradigm to promote the acceptance, performance, and viability of AI technologies in agriculture. In addition to its technical performance, the trust, engagement, and sustainability of farmers with AI-driven systems are influenced by user experience factors including usability, transparency, cultural relevance, and accessibility (Wu and Zhong, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Omotayo et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). The literature on AI solutions reviewed by Scopus in the recent past has a growing focus on the need to design AI solutions, basing them on the needs, situations, and abilities of farmers to facilitate a comprehensive and equitable transformation of agriculture.\u003c/p\u003e \u003cp\u003eCo-Design and Participatory Development.\u003c/p\u003e \u003cp\u003eThe extensive use of co-design and participatory development methods is also accepted as a key to the harmonization of AI technologies and real-world farming. Instead of viewing farmers as passive end-users, participatory models engage them during the design, testing, and refinement phases of AI systems, which guarantees that the tools are designed to capture local agronomic knowledge and decision-making (Ryan et al., \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Mana et al., \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Precision agriculture and AIoT application evidence indicates that participatory engagement can increase system relevance, lower technological change resistance, and perceived usefulness among farmers (Bayar et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Mamabolo et al., \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe human-centered design has especially been found to work in smallholder and resource-constrained environments, where the standardized AI solutions fail to capture the variety of cropping systems, the patchwork nature of landholdings, and socio-economic limits. Research points out that participatory pilots and feedback loops enable adapting the AI-based disease detectors, irrigation scheduling, and advisory systems to the local conditions, thus enhancing adoption results (Su et al., \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2026\u003c/span\u003e; High et al., \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Besides, mediating the participation of an AI in development wherein technical concepts are transformed into practice-based solutions is the role of extension agents and farmer organizations (Lee et al., \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Tran Cao Uy et al., \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2024\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eExplainable AI and Localized Interfaces.\u003c/p\u003e \u003cp\u003eExplainable AI (XAI) is regarded more and more as a foundation of trust and acceptance of agricultural AI systems by users. Complex black-box models are highly accurate, but with little transparency and interpretability, they may not be able to win the trust of farmers (Gardezi et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Majdalawieh et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). The literature also highlights that farmers will tend to trust AI recommendations more when the systems have understandable explanations of the outputs, i.e., visual indicators, rule-based logic, or simplified confidence scores (Aijaz et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Wu and Zhong, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eLocalized user interfaces also improve usability by tailoring AI outputs in accordance with the cognitive frames and realities in the work of farmers. Demonstrated to be more efficient as far as the decision-making process is concerned and to create less cognitive load, context-specific dashboards, mobile-based notifications, and visual cues have been adapted to local crops and practices (Padhiary et al., \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Rashid et al., \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Research also mentions that explainability is especially significant in high-stakes decisions, like the choice of pesticides or the time of irrigation, where farmers require to know the reasoning behind AI-based recommendations in order to reduce the perceived risks (Omotayo et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; \u0026Ouml;zoğul, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eAccessibility, Cultural Relevancy, and Language.\u003c/p\u003e \u003cp\u003eThe issue of access to AI technologies in the Global South revolves around language and cultural relevance. Studies always show that AI systems optimized on the main languages of dominance or belonging to the global community impose an obstacle to farmers with low levels of formal education or literacy (Daum, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Dittmer et al., \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Voice-based advisory systems, local-language interfaces, and icon-based visualizations are much more effective in understanding and interacting with the application, especially among smallholder farmers (Nguyen et al., \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Sharma et al., 2025).\u003c/p\u003e \u003cp\u003eCultural relevance goes beyond language and encompasses cultural correspondence to local agricultural practices, perceptions of risk and social constructs. Research stresses that the technological savvy of the AI tools lacks acceptance in many instances when indigenous knowledge systems or even known practices are ignored (Atasoy, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Hiywotu, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). On the other hand, culturally adaptive AI systems, i.e., systems that incorporate local heuristics and knowledge of farmers, are less likely to be perceived as disruptive and have more chances of being perceived as supportive (High et al., \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Lankamo et al., \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe accessibility also includes affordability, compatibility with the devices, and infrastructure limitations. Smaller AI models, offline experience and mobile-first designs are also being encouraged in order to make sure that they can be used in areas with weak connection and lack of hardware access (Saiz et al., \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2026\u003c/span\u003e; Abdulraheem et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2026\u003c/span\u003e). Collectively, these anthropocentric considerations highlight the fact that the successful implementation of AI in agriculture is not only a factor of the functionality of the algorithms but also the degree of intelligibility, applicability, and accessibility of the systems to different farming groups.\u003c/p\u003e \u003c/div\u003e"},{"header":"6. Barriers and Enablers for AI Implementation","content":"\u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003e6.1 Technical and Infrastructure Barriers\u003c/h2\u003e \u003cp\u003eAlthough artificial intelligence (AI) applications are increasingly becoming mature in agriculture, its mass application to the enterprise is still limited due to the ongoing technical and infrastructural challenges. Such challenges are especially acute regarding developing agricultural systems and smallholder agriculture, in which digital ecosystems have remained fragmented and unevenly developed.\u003c/p\u003e \u003cp\u003eConnectivity Gaps\u003c/p\u003e \u003cp\u003eDependable online connectivity is a pre-requisite of AI-supported farming technologies, such as cloud-based decision support systems, AIoT systems, UAV analytics as well as real time sensor networks. Nevertheless, there are still significant connectivity differences between urban and rural areas, which restricts the functionality and scalability of precision farming through AI. Poor broadband connectivity, unreliable mobile signals, and excessively expensive data transfer create barriers in real-time data derived and model implementation, particularly applications like smart irrigation, illness overview, and self-operated machinery (Bayar et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Rashid et al., \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Research articles are consistent in reporting that lack of quality digital infrastructure limits access to AI services by farmers and reduces feedback loops needed by adaptive learning systems (Saiz et al., \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2026\u003c/span\u003e; Ishore et al., \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Such gaps increase the gap in digital inequalities and diminish the benefits in productivity and sustainability that AI-based innovations in agriculture promise (Wu and Zhong, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Ahmad et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eData Sparingness and Data Quality.\u003c/p\u003e \u003cp\u003eThe basis of AI systems in agriculture is data-intensive, as large amounts of high-quality, context-dependent data are needed to train, validate and continue improving. Nevertheless, the lack of data is still a severe bottleneck especially in places where smallholder farming is predominant and there are mixed agro-ecological contexts. The weaknesses of AI models include limited access to labeled datasets, uneven data collection, temporal and spatial gaps in data, which deteriorate the strength and generalizability of AI models (Aijaz et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Filippi et al., \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Besides, the quality of data, including sensor noise, missing data, biased sample, and non-standardization, lowers the predictive accuracy and confidence of farmers in AI results (Majdalawieh et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Gupta and Kumar Pal, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). A number of surveys point out that the lack of integration of local agronomic insights and ground-truth validation further limits the model transferability between regions and cropping systems (Mana et al., \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Upadhyay et al., \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). To overcome these issues, it is necessary to invest in data governance systems, participatory data collection, and interoperable agricultural data standards.\u003c/p\u003e \u003cp\u003eInteroperability Challenges\u003c/p\u003e \u003cp\u003eThe other major technical obstacle to successful AI implementation in agriculture is interoperability. The existing state of the digital agriculture sector can be described as the increase in the number of proprietary platforms, heterogeneous sensors, and vendor-specific software ecosystems that are not always compatible. Consequently, the combination of data streams of IoT devices, UAVs, farm machinery, and external data is still complicated and expensive (Padhiary et al., \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Rashid et al., \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). The lack of interoperability hinders the transfer of data and the creation of farm-scale decision support devices (Omotayo et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Ryan et al., \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Moreover, the fact that there are no common principles of data formats, communication protocols and the implementation of the AI models complicates the scalability of the systems and their long-term sustainability (\u0026Ouml;zoğul, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Abdulraheem et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2026\u003c/span\u003e). Researchers point out that interoperability barriers will only be overcome through a concerted effort by technology vendors, policy makers and research institutions to ensure open architectures, standardization and modular system design (Wu and Zhong, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Vijayakumar et al., 2025).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003e6.2 Institutional and Economic Barriers.\u003c/h2\u003e \u003cp\u003eAlthough artificial intelligence (AI) technologies have a high potential of increasing productivity, sustainability and risks management in agriculture, economic and institutional barriers are also major limiting factors to their diffusion. The smallholder and resource-constrained farmers are disproportionately impacted by these barriers and result in unequal access to AI-enabled innovations and slows system-wide change.\u003c/p\u003e \u003cp\u003eCost of AI Tools\u003c/p\u003e \u003cp\u003eThe prohibitive initial and maintenance expenses that are involved in AI-agricultural technologies continue to be one of the biggest obstacles to adoption. Small and medium-scale farmers may not be financially able to invest in sensors, UAVs, smart machinery, cloud subscriptions, and data analytics platforms. Besides the purchase of the hardware, the maintenance of the system, updates of the software, data storage, technical support, among others, make this total cost of ownership even greater (Aijaz et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Padhiary et al., \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Empirical and review research has shown that perceived economic risk and uncertain payoff on investment deter farmers to use AI solutions, which is more evident when prices are volatile and uncertain about the climate (Mamabolo et al., \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; \u0026Ouml;zoğul, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Even though AIoT and automation technologies have the potential to achieve long-term efficiency benefits, their cost is firmly linked to the economies of scale, which supports asymmetries in adoption between large commercial farms and smallholders (Bayar et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Gupta and Kumar Pal, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eAbsence of Insurance and Credit Intersection.\u003c/p\u003e \u003cp\u003eThe minor integration of the AI technologies into the agricultural credit and insurance frameworks is a severe institutional chokepoint. Although AI-controlled analytics has potential, the ability to enhance risk evaluation, yield forecasting, and loss verification, they are not often offered in formal financial services available to farmers (Omotayo et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Wu and Zhong, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Most organizations do not provide new types of credit products, which acknowledge digital assets or data-based performance measurements, limiting farmers when making AI investments (Sood et al., \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Hiywotu, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Equally, the lack of AI-based crop insurances schemes lowers the use of technology incentives because farmers will not be under the protection of sufficient safety nets against climatic risks and market risks (Saiz et al., \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2026\u003c/span\u003e; Vijayakumar et al., 2025). Researchers stress that to reduce the risks of adoption and increase economic feasibility of precision agriculture, there is a need to have greater alignment between AI systems, financial institutions, and insurance companies (Ryan et al., \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Ahmad et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eDisjointed Policy Ecosystems.\u003c/p\u003e \u003cp\u003eScattered and inconsistent policy landscapes also restrict the scalability of AI in agriculture. Digital agriculture strategies in most countries are spread over various ministries and agencies, which leads to overlapping mandates, regulatory confusion, and low coordination between policies of innovation, data governance, and rural development (Ahmad et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Omotayo et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). The absence of explicit policies on data possession, privacy, interoperability and deployment ethics of AI poses uncertainties to both the technology providers and the end-users (Atasoy, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; \u0026Ouml;zoğul, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). In addition, the lack of governmental funding of digital public goods, including open datasets, advisory services, and AI services that are based on extensions, decreases the institutional support of inclusion in adoption (Daum, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Dittmer et al., \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). They indicate that coherent policy frameworks, as well as incentive-based programs and public- private partnerships, have a key role in building trust, lowering transaction costs and facilitating sustainable integration of AI technologies in agricultural value chains (Wu and Zhong, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Vijayakumar et al., 2025).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec23\" class=\"Section2\"\u003e \u003ch2\u003e6.3 Enablers and Success Factors\u003c/h2\u003e \u003cp\u003eA combination of institutional, technological, and policy-related enablers is significant in the success of artificial intelligence (AI) adoption and scaling in agriculture. Recent scopus-indexed literature points out that in addition to technological preparedness, models of service delivery, collaborative governance, digital advisory systems, and enabling public policies have a decisive role to play in the determination of impact and sustainability.\u003c/p\u003e \u003cp\u003eAI-as-a-Service Models\u003c/p\u003e \u003cp\u003eAI-as-a-Service has become an important facilitator of reduced barriers of entry to advanced digital technologies in agriculture. Rather than demanding farmers to invest in expensive hardware, software, and skilled employees, AIaaS enables the utilization of analytics and decision-support applications and predictive models via cloud-based technologies on subscription or a pay-per-use basis. Research notes that AIaaS allows deploying precise farming solutions (including yield prediction, disease detection, and optimization of irrigation) on a larger scale especially to small and medium-scale farmers (Aijaz et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Mana et al., \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2024\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eAIaaS is also highly associated with AIoT systems, in which sensor data of the fields, UAVs, and smart equipment are centrally processed to show real-time insights, thus increasing the efficiency of the resources and operational decision-making (Bayar et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Rashid et al., \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). In addition, service-oriented AI provision eliminates risks associated with system maintenance, model updates, and data security, which tend to be mentioned as obstacles to the use of AI on the farm (Omotayo et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Wu and Zhong, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003ePublic-Private Partnerships (PPP).\u003c/p\u003e \u003cp\u003eThe concept of public-private partnerships is commonly identified as potential success factors in fast-tracking the process of AI innovation and diffusion into the agricultural sector. PPPs help leverage the strengths of data analytics, platform development, and commercialization (that are part of the capabilities of the private sector) with the objectives of food security, sustainability, and inclusion (that are part of the goals of the public sector) (Ahmad et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Ryan et al., \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eAccording to evidence PPPs are an important driver of the piloting of AI solutions, the development of a common data infrastructure, and alignment to local agronomic and socio-economic environments (Majdalawieh et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Gupta and Kumar Pal, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Moreover, PPPs can be used to build trust in farmers by integrating AI tools into publicly approved programs and extension systems and, therefore, decrease farmers' distrust of individual digital technologies (Gardezi et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Omotayo et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eDigital Extension Systems\u003c/p\u003e \u003cp\u003eDigital extension systems constitute a radical facilitator of reaching out to farmers via AI technologies. AI-based advisory systems, mobile apps, and decision-support applications can help reach more customers with its traditional extension services and provide customized recommendations in real-time, location-based, and personalized (High et al., \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Tran Cao Uy et al., \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2024\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eLiterature highlights the importance of digital extension systems in increasing the ability of farmers to understand AI outputs and enhancing perceived usefulness and intention to adopt AI-driven practices (Lee et al., \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Wu and Zhong, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Combining AI and digital extension also facilitates continual learning based on feedback loops, peer interaction, and data-driven demonstrations that are especially useful in smallholder and resource-limited settings (Dittmer et al., \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Saiz et al., \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2026\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eOpen Data Platforms and Government Subsidies.\u003c/p\u003e \u003cp\u003eThe essential success factors for the equitable use of AI in agriculture are government assistance, such as subsidies and open data programs. Digital infrastructure, smart sensors, and AI-enabled machines subsidies lower financial risks and encourage early adoption, particularly in smallholder farmers (Hiywotu, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Vijayakumar et al., 2025).\u003c/p\u003e \u003cp\u003eAgri-data websites such as weather, soil, crop conditions, and market data are also recurrently mentioned as competition-free inputs to train powerful AI models and develop innovation ecosystems (Wu and Zhong, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Atasoy, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Both transparency and interoperability and innovation are facilitated by publicly available datasets through allowing startups, researchers, and extension agencies to develop AI solutions more specific to local needs (Aijaz et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; \u0026Ouml;zoğul, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Combined, subsidies and open data policies can empower the facilitation of sustainable, inclusive, and ethically based AI implementation in the agricultural sector (Omotayo et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e"},{"header":"7. An Integrated Framework for AI Adoption in Smallholder Agriculture","content":"\u003cdiv id=\"Sec25\" class=\"Section2\"\u003e\n \u003ch2\u003e7.1 Smallholder-Centric AI Adoption Framework\u003c/h2\u003e\n \u003cp\u003eI. Technology Layer: Accuracy, Scalability, and Explainability.\u003c/p\u003e\n \u003cp\u003eTechnology Layer smallholder agriculture AI adoption focuses on AI tool and model accuracy, scalability and explainability. The precision makes smallholders with limited resources reliable in crop prediction, detection of pests, and optimization of irrigation, which is essential (Aijaz et al., \u003cspan class=\"CitationRef\"\u003e2025\u003c/span\u003e; Majdalawieh et al., \u003cspan class=\"CitationRef\"\u003e2025\u003c/span\u003e). Scalability enables the application of AI-driven IoT devices, UAVs, and AIoT solutions to different scales of farms to make more farmers use the precision farming practices (Bayar et al., \u003cspan class=\"CitationRef\"\u003e2025\u003c/span\u003e; Rashid et al., \u003cspan class=\"CitationRef\"\u003e2025\u003c/span\u003e). Explainability would allow farmers to be familiar with AI recommendations and more likely to trust and implement the tools, overcoming the challenges caused by black-box AI systems (Wu and Zhong, \u003cspan class=\"CitationRef\"\u003e2025\u003c/span\u003e; Mana et al., \u003cspan class=\"CitationRef\"\u003e2024\u003c/span\u003e). Combined, these technological factors form the basis for effective and acceptable AI systems for smallholder farmers.\u003c/p\u003e\n \u003cp\u003eII. Human Layer: Perceived Value, Trust, and Skills.\u003c/p\u003e\n \u003cp\u003eThe Human Layer pays attention to the socio-cognitive factors that affect the adoption of AI. The confidence in AI systems is a decisive factor because smallholders will be more willing to use solutions that can be seen as reliable and transparent (Gardezi et al., \u003cspan class=\"CitationRef\"\u003e2023\u003c/span\u003e; Su et al., \u003cspan class=\"CitationRef\"\u003e2026\u003c/span\u003e). To use, interpret, and gain access to AI technologies, one will require sufficient skills, such as digital literacy and knowledge of AI-assisted farming practices (Daum, \u003cspan class=\"CitationRef\"\u003e2025\u003c/span\u003e; High et al., \u003cspan class=\"CitationRef\"\u003e2025\u003c/span\u003e). Last but not least, the perceived value of AI systems, the perception of farmers about the positive effect of the systems on productivity, income, or sustainability, is a strong determinant of adoption (Nguyen et al., \u003cspan class=\"CitationRef\"\u003e2023\u003c/span\u003e; Sood et al., \u003cspan class=\"CitationRef\"\u003e2022\u003c/span\u003e). Any initiatives that address these elements will ensure more involvement and higher chances of the continued adoption of AI among smallholders.\u003c/p\u003e\n \u003cp\u003eIII. System Layer: Policy, infrastructure, and Markets.\u003c/p\u003e\n \u003cp\u003eSystem Layer is an enabling environment to adopt AI. Policies on the national and regional levels, including subsidies, assistance in the development of innovations, and regulation, become favorable to the implementation of AI (Omotayo et al., \u003cspan class=\"CitationRef\"\u003e2025\u003c/span\u003e; Ahmad et al., \u003cspan class=\"CitationRef\"\u003e2025\u003c/span\u003e). The implementation of AI technologies on a large scale and inclusion of the smallholders requires adequate infrastructure, such as power, internet access, and digital services (Saiz et al., \u003cspan class=\"CitationRef\"\u003e2026\u003c/span\u003e; Tran Cao Uy et al., \u003cspan class=\"CitationRef\"\u003e2024\u003c/span\u003e). Lastly, economic incentives to smallholders to use AI are reinforced by the presence of market mechanisms that offer fair prices, access to inputs, and trading platforms of selling outputs (Hiywotu, \u003cspan class=\"CitationRef\"\u003e2025\u003c/span\u003e; Erickson and Fausti, 2021). The combination of these systemic components will provide the adequate support of technological and human capabilities of the agricultural ecosystem.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec26\" class=\"Section2\"\u003e\n \u003ch2\u003e7.2 AI Innovation to Productivity Gains Pathways.\u003c/h2\u003e\n \u003cp\u003eThe channels by which AI innovations can be converted into productivity increase are mediated by adoption processes and cyclic engagement between smallholder farmers and technology developers.\u003c/p\u003e\n \u003cp\u003eMediating AI Impact through Adoption.\u003c/p\u003e\n \u003cp\u003eAdoption is an important intermediary between AI technologies and real gains in agricultural productivity. Although AI-based technologies, including precision irrigation, pest detection, and yield predictions, have a high technical potential, their application will rely on the adoption by smallholders (Aijaz et al., \u003cspan class=\"CitationRef\"\u003e2025\u003c/span\u003e; Bayar et al., \u003cspan class=\"CitationRef\"\u003e2025\u003c/span\u003e). The willingness of farmers to use AI systems, perceptions of usefulness, and competency directly affect the extent of AI interventions generating demonstrable results in increase in crop yields, resource management, and revenue (Nguyen et al., \u003cspan class=\"CitationRef\"\u003e2023\u003c/span\u003e; Sood et al., \u003cspan class=\"CitationRef\"\u003e2022\u003c/span\u003e). Moreover, it depends on the availability and convenience of technology, such as smartphone applications, IoT sensors, and automated farm machines (Padhiary et al., \u003cspan class=\"CitationRef\"\u003e2024\u003c/span\u003e; Rashid et al., \u003cspan class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e\n \u003cp\u003eFeedback between Technology Developers and Farmers.\u003c/p\u003e\n \u003cp\u003eThe proper implementation of AI depends on the feedback loops that would enable further learning between farmers and developers. Smallholders can deliver current information and contextual data regarding the state of the soil, pest behavior, and crop data that can be used to optimize machine learning models and enhance AI decision-making (Wu and Zhong, \u003cspan class=\"CitationRef\"\u003e2025\u003c/span\u003e; Upadhyay et al., \u003cspan class=\"CitationRef\"\u003e2025\u003c/span\u003e). Human-oriented AI tools and participatory design practices are used to make sure that the technologies can adapt to the needs of farmers, and the cycle of innovation-adoption-productivity improvement develops (Su et al., \u003cspan class=\"CitationRef\"\u003e2026\u003c/span\u003e; High et al., \u003cspan class=\"CitationRef\"\u003e2025\u003c/span\u003e). Such iterative feedback mechanisms prove especially useful when dealing with resource-constrained settings where generic AI solutions are prone to fail when they do not adapt locally (Omotayo et al., \u003cspan class=\"CitationRef\"\u003e2025\u003c/span\u003e; Mana et al., \u003cspan class=\"CitationRef\"\u003e2024\u003c/span\u003e).\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec27\" class=\"Section2\"\u003e\n \u003ch2\u003e7.3 Comparison to Existing Models.\u003c/h2\u003e\n \u003cp\u003eThe Framework Extension of Agriculture 4.0 / 5.0 Models.\u003c/p\u003e\n \u003cp\u003eAgriculture 4.0 and 5.0 models focus on automation, IoT connectivity, and AI-driven optimization (Baladraf et al., \u003cspan class=\"CitationRef\"\u003e2025\u003c/span\u003e; Gupta et al., 2025), whereas the Smallholder-Centric AI Adoption Framework directly incorporates technological, human, and system layers with the view of considering the smallholder-specific constraints. In contrast to the generic smart agriculture models, this framework emphasizes the importance of the element of trust, skills, and perceived value in adoption, as well as facilitating policies, infrastructure, and market support (Tran Cao Uy et al., \u003cspan class=\"CitationRef\"\u003e2024\u003c/span\u003e; Daum, \u003cspan class=\"CitationRef\"\u003e2025\u003c/span\u003e). This is in the holistic approach such that the AI innovations are not only of an advanced technical nature, but also affordable and socially accessible to the small scale farmers.\u003c/p\u003e\n \u003cp\u003eValue Addition to the Developing-Country Situations.\u003c/p\u003e\n \u003cp\u003eThe framework also brings considerable value to the context of the developing countries by considering socio-technical and economic challenges that restrict the adoption of AI. It also focuses on participatory design, integration of local knowledge and feedback loops to make AI interventions specific to local crop systems, resource availability, and market structures (Hiywotu, \u003cspan class=\"CitationRef\"\u003e2025\u003c/span\u003e; Erickson and Fausti, 2021). Also the framework takes into consideration policy and infrastructure gaps that are typical of low- and middle-income countries, building a more robust and inclusive channel between the initiation of AI innovations and productivity improvements (Omotayo et al., \u003cspan class=\"CitationRef\"\u003e2025\u003c/span\u003e; Ahmad et al., \u003cspan class=\"CitationRef\"\u003e2025\u003c/span\u003e). In general, it offers a viable blueprint of how to transform the latest AI technologies into viable productivity and income gains to the smallholders.\u003c/p\u003e\n\u003c/div\u003e"},{"header":"8. Implications of Policy, Design, and Research.","content":"\u003cp\u003eThe introduction of AI in smallholder agriculture must be a collective effort in policy, technology development, and research priorities. Stakeholders can maximize the benefits of AI to achieve sustainable agricultural productivity by solving socio-technical obstacles and providing inclusivity.\u003c/p\u003e \u003cdiv id=\"Sec29\" class=\"Section2\"\u003e \u003ch2\u003e8.1 Policy Implications.\u003c/h2\u003e \u003cp\u003eDigital Infrastructure Investment: A successful application of AI depends on the good digital infrastructure, such as access to reliable internet connectivity, wireless networks, and mobile platforms, which are typically lacking in rural areas of the developing world (Tran Cao Uy et al., \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Daum, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Digital infrastructure investments make it possible to collect data in time, use it in precision farming, and provide better access to the market to smallholders (Ahmad et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Wu and Zhong, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eInclusive AI Governance: Policymakers should come up with governance systems that are inclusive to all stakeholders and safeguard the smallholder against any risks, such as misuse of data and algorithm bias (Omotayo et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Abdulraheem et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2026\u003c/span\u003e). Inclusive AI governance leads to accountability, transparency and local stakeholder involvement in AI policy making.\u003c/p\u003e \u003cp\u003eSmallholder-Centric AI Solutions: National and regional solutions must target smallholder farmers through subsidies of AI-enabled solutions, extension, and combining AI solutions with local agricultural knowledge systems (Hiywotu, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; High et al., \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Specific policies will help to close the technology gap and help to increase productivity by providing benefits to resource-constrained farmers.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec30\" class=\"Section2\"\u003e \u003ch2\u003e8.2 Technology Developers Implications.\u003c/h2\u003e \u003cp\u003eDesign to Low-Resource Environments: AI systems must be designed to run well in low-resource environments, being intermittently connected, with limited access to hardware, and a variety of cropping conditions (Padhiary et al., \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Bayar et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Mobile applications and smallweight AI tools and offline capabilities are necessary to make them usable by smallholders.\u003c/p\u003e \u003cp\u003eReliable and understandable AI: Trust is a significant factor in the use of AI. Explainable AI models should be a priority among developers that give unambiguous, practical advice and allow farmers to interpret results of the algorithms (Gardezi et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Su et al., \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2026\u003c/span\u003e). Open AI systems will lessen non-cooperation and build more acceptance among smallholders.\u003c/p\u003e \u003cp\u003eCo-Creation with Farmers: Participatory design solutions, in which the farm is included in the process of creating the AI, enhance relevance, usability, and adoption rates (High et al., \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Mana et al., \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Developers and farmers can also provide feedback on AI tools to refine them over time through feedback loops to align with local agricultural practices and social-economic settings (Wu and Zhong, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Upadhyay and others, 2025).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec31\" class=\"Section2\"\u003e \u003ch2\u003e8.3 Future Research Implications.\u003c/h2\u003e \u003cp\u003eLongitudinal Impact Studies: Longitudinal studies that could evaluate long-term impacts of AI adoption on productivity, income, and sustainability are required (Aijaz et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Erickson and Fausti, 2021). Knowledge of temporal impacts will help policymakers and developers to come up with more efficient interventions.\u003c/p\u003e \u003cp\u003eGender and Equity Dimensions: To make sure that AI adoption advantages all demographic groups, such as women farmers and marginalized communities, future studies are required to examine the impact of AI adoption on the specified groups (Nguyen et al., \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Sood et al., \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2022\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eAI Ethics in Rural Areas: Ethical concerns, such as data privacy, algorithmic fairness and informed consent have not been studied in rural agricultural areas (Omotayo et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Abdulraheem et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2026\u003c/span\u003e). Studies ought to examine the culturally appropriate systems to protect the smallholders and at the same time enhance AI development.\u003c/p\u003e \u003c/div\u003e"},{"header":"9. Conclusion","content":"\u003cp\u003eThe paper describes the importance of artificial intelligence in terms of improving productivity, sustainability, and resilience in smallholder farming. The offered framework unites technological, human, and systemic aspects, which contributes to a full-scale explanation of the ways of AI adoption and its possible consequences.\u003c/p\u003e \u003cp\u003eSummary of Key Findings: The study has established that the implementation of AI is not only contingent upon technological, i.e., accuracy, scalability, and explainability, but also upon human factors, i.e. trust, skills and perceived value, and system-level enablers, i.e. policy, infrastructure and market access (Aijaz et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Bayar et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Wu and Zhong, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). The results show that feedback loops exist between farmers and technology developers in the mediation of adoption, and co-creation and participatory design are the most important elements of successful adoption (High et al., \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Su et al., \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2026\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eRestatement of Research Objectives: The paper has approached the research goals of comprehending the adoption of AI through the perspective of smallholders, visualizing the routes to productivity improvement, and placing the paradigm in the context of current Agriculture 4.0 and 5.0 paradigms. The findings reveal that the suggested framework expands existing models by including socio-technical and equity aspects, which are especially critical in the framework of developing countries where resource-related limitations and digital divides still exist (Hiywotu, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Tran Cao Uy et al., \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2024\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe stress on Human-Centered, System-Aware AI: The research confirms that AI cannot be a completely technical solution. To make sure that it will be meaningfully adopted and used, it is imperative to incorporate human-centered design, explainability, and trust-building. Similarly, system-sensitive interventions such as enabling policies, market connections, and infrastructure are needed to implement AI inventions as actual productivity (Omotayo et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Ahmad et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Mana et al., \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2024\u003c/span\u003e).\u003c/p\u003e \u003cp\u003ePlacing AI as an Enabler, not a Standalone Solution: AI should be regarded as an enabler of smallholder productivity, and not a solution in and of itself. It relies on the incorporation of local knowledge, the involvement of farmers, and institutional support. AI can be used in a targeted way to change the smallholder agriculture sector through achieving sustainability, resilience, and equitable growth (Erickson and Fausti, 2021; Abdulraheem et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2026\u003c/span\u003e; Padhiary et al., \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2024\u003c/span\u003e).\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eClinical Trial Number\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eClinical trial number: not applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics, Consent to Participate, and Consent to Publish\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eEthics, consent to participate, and consent to publish: not applicable.\u003cbr\u003e\u0026nbsp;This study does not involve human participants or animals and therefore does not require ethical approval or consent.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eAbdulraheem, M. I., Khan, Z. H., \u0026amp; Hu, J., et al. (2026). AgriTech: Prospects and challenges of robotics, artificial intelligence \u0026amp; 5G technologies in agriculture. \u003cem\u003eJournal of Ambient Intelligence and Smart Environments.\u003c/em\u003e https://doi.org/10.1177/18761364251409357\u003c/li\u003e\n\u003cli\u003eAhmad, A., Murad, M., Utra, U., Yussof, N. S., Fazial, F. F., \u0026amp; Lazim, A. (2025). Transforming food systems with artificial intelligence: Challenges, governance, and pathways for sustainable integration. \u003cem\u003eFood Reviews International.\u003c/em\u003e https://doi.org/10.1080/87559129.2025.2585134\u003c/li\u003e\n\u003cli\u003eAjith, S., Vijayakumar, S., \u0026amp; Elakkiya, N. (2025). Yield prediction, pest and disease diagnosis, soil fertility mapping, precision irrigation scheduling, and food quality assessment using machine learning and deep learning algorithms. \u003cem\u003eDiscover Food, 5,\u003c/em\u003e 67. https://doi.org/10.1007/s44187-025-00338-1\u003c/li\u003e\n\u003cli\u003eAijaz, N., Lan, H., Raza, T., Yaqub, M., Iqbal, R., \u0026amp; Pathan, M. S. (2025). Artificial intelligence in agriculture: Advancing crop productivity and sustainability. \u003cem\u003eJournal of Agriculture and Food Research, 20,\u003c/em\u003e 101762. https://doi.org/10.1016/j.jafr.2025.101762\u003c/li\u003e\n\u003cli\u003eAlmazmomi, N. K. (2025). Artificial intelligence-driven blockchain and Internet of Things framework for secure data management in precision agriculture. \u003cem\u003eJournal of High Speed Networks, 31\u003c/em\u003e(3). https://doi.org/10.1177/09266801251331816\u003c/li\u003e\n\u003cli\u003eAtasoy, Y. (2025). Agriculture by algorithm: Big data, digitalization, and biotechnology under climate change. \u003cem\u003eScience, Technology, \u0026amp; Human Values.\u003c/em\u003e https://doi.org/10.1177/01622439251321233\u003c/li\u003e\n\u003cli\u003eBaladraf, T. T., Marimin, Hermadi, I., \u0026amp; Ismayana, A. (2025). Review of Agriculture 5.0: Artificial intelligence, IoT, and machine learning. \u003cem\u003eJournal of Agricultural \u0026amp; Food Information.\u003c/em\u003e https://doi.org/10.1080/10496505.2025.2583361\u003c/li\u003e\n\u003cli\u003eBayar, J., Ali, N., Cao, Z., Ren, Y., \u0026amp; Dong, Y. (2025). Artificial intelligence of things (AIoT) for precision agriculture: Applications in smart irrigation, nutrient and disease management. \u003cem\u003eSmart Agricultural Technology, 12,\u003c/em\u003e 101629. https://doi.org/10.1016/j.atech.2025.101629\u003c/li\u003e\n\u003cli\u003eCardak, M. C., Golec, M., \u0026amp; Gill, S. S. (2025). FarmTechAI: Artificial intelligence-based modern farmer management system. In A. Kumar, J. P. Verma, \u0026amp; R. Jain (Eds.), \u003cem\u003eEmerging smart agricultural practices using artificial intelligence\u003c/em\u003e (Chap. 12). Wiley. https://doi.org/10.1002/9781394274277.ch12\u003c/li\u003e\n\u003cli\u003eClay, D. E., Brugler, S., \u0026amp; Joshi, B. (2024). Will artificial intelligence and machine learning change agriculture: A special issue. \u003cem\u003eAgronomy Journal, 116\u003c/em\u003e(3). https://doi.org/10.1002/agj2.21555\u003c/li\u003e\n\u003cli\u003eDaum, T. (2025). Digitalization and skills in agriculture. \u003cem\u003eOutlook on Agriculture, 54\u003c/em\u003e(2). https://doi.org/10.1177/00307270251336474\u003c/li\u003e\n\u003cli\u003eDittmer, K. M., Burns, S., \u0026amp; Wollenberg, E. (2025). Digital tool innovations for smallholder inclusion. \u003cem\u003eOutlook on Agriculture, 54\u003c/em\u003e(3). https://doi.org/10.1177/00307270251331644\u003c/li\u003e\n\u003cli\u003eFilippi, P., Han, S. Y., \u0026amp; Bishop, T. F. A. (2025). On crop yield modelling, predicting, and forecasting and addressing the common issues in published studies. \u003cem\u003ePrecision Agriculture, 26,\u003c/em\u003e 8. https://doi.org/10.1007/s11119-024-10212-2\u003c/li\u003e\n\u003cli\u003eGardezi, M., Joshi, B., Rizzo, D. M., Ryan, M., Prutzer, E., Brugler, S., \u0026amp; Dadkhah, A. (2023). Artificial intelligence in farming: Challenges and opportunities for building trust. \u003cem\u003eAgronomy Journal, 115\u003c/em\u003e(4). https://doi.org/10.1002/agj2.21353\u003c/li\u003e\n\u003cli\u003eGupta, G., \u0026amp; Kumar Pal, S. (2025). Applications of artificial intelligence in precision agriculture. \u003cem\u003eDiscover Agriculture, 3,\u003c/em\u003e 61. https://doi.org/10.1007/s44279-025-00220-9\u003c/li\u003e\n\u003cli\u003eHiywotu, A. M. (2025). Advancing sustainable agriculture for goal 2: Zero hunger \u0026ndash; A comprehensive overview of practices, policies, and technologies. \u003cem\u003eAgroecology and Sustainable Food Systems, 49\u003c/em\u003e(7), 1027\u0026ndash;1055. https://doi.org/10.1080/21683565.2025.2451344\u003c/li\u003e\n\u003cli\u003eHigh, C., Singh, N., \u0026amp; Nemes, G. (2025). Artificial intelligence for agricultural extension: Supporting transformative learning among smallholder farmers. \u003cem\u003eJournal of Development Policy and Practice, 11\u003c/em\u003e(1). https://doi.org/10.1177/2455133325134522\u003c/li\u003e\n\u003cli\u003eIshore, B., Bhargav, A. M., Patel, S. K., Kumar, V., Sinha, J., \u0026amp; Kumar, S. (2025). Advancing smart agriculture: How IoT and 5G technologies enhance precision farming and sustainability. \u003cem\u003eInternational Journal of Communication Systems.\u003c/em\u003e https://doi.org/10.1002/dac.70310\u003c/li\u003e\n\u003cli\u003eKuradusenge, M., et al. (2024). SMART-CYPS: An intelligent IoT and machine learning\u0026ndash;powered crop yield prediction system for food security. \u003cem\u003eDiscover Internet of Things, 4,\u003c/em\u003e 23. https://doi.org/10.1007/s43926-024-00079-0\u003c/li\u003e\n\u003cli\u003eLankamo, A. A., Ramalingam, D., \u0026amp; Dira, S. J., et al. (2025). Adoption dynamics of climate-smart agriculture. \u003cem\u003eSAGE Open.\u003c/em\u003e https://doi.org/10.1177/21582440251343674\u003c/li\u003e\n\u003cli\u003eLee, C.-L., Strong, R., Briers, G., Murphrey, T., Rajan, N., \u0026amp; Rampold, S. (2024). Factors predicting innovation-decisions: The effects of performance expectancy, social influence, and facilitating conditions on U.S. Extension\u0026rsquo;s promotion of precision agriculture technologies. \u003cem\u003eNJAS: Impact in Agricultural and Life Sciences, 96\u003c/em\u003e(1). https://doi.org/10.1080/27685241.2024.2420111\u003c/li\u003e\n\u003cli\u003eMahale, Y., et al. (2024). Crop recommendation and forecasting system for Maharashtra using machine learning with LSTM. \u003cem\u003eDiscover Sustainability, 5,\u003c/em\u003e 134. https://doi.org/10.1007/s43621-024-00292-5\u003c/li\u003e\n\u003cli\u003eMamabolo, E., Mashala, M. J., Mugari, E., Mogale, T. E., Mathebula, N., Mabitsela, K., \u0026amp; Ayisi, K. K. (2025). Application of precision agriculture technologies for crop protection and soil health. \u003cem\u003eSmart Agricultural Technology, 12,\u003c/em\u003e 101270. https://doi.org/10.1016/j.atech.2025.101270\u003c/li\u003e\n\u003cli\u003eMajdalawieh, M., Martins, C., Radi, M., Alaraj, M., \u0026amp; Khan, S. (2025). Precision agriculture in the age of AI: A systematic review of machine learning methods for crop disease detection. \u003cem\u003eSmart Agricultural Technology, 12,\u003c/em\u003e 101491. https://doi.org/10.1016/j.atech.2025.101491\u003c/li\u003e\n\u003cli\u003eMana, A. A., Allouhi, A., Hamrani, A., Rehman, S., El Jamaoui, I., \u0026amp; Jayachandran, K. (2024). Sustainable AI-based production agriculture: Exploring AI applications and implications in agricultural practices. \u003cem\u003eSmart Agricultural Technology, 7,\u003c/em\u003e 100416. https://doi.org/10.1016/j.atech.2024.100416\u003c/li\u003e\n\u003cli\u003eMinhans, K., Sharma, S., Sheikh, I., Alhewairini, S. S., \u0026amp; Sayyed, R. (2025). Artificial intelligence and plant disease management: An agro-innovative approach. \u003cem\u003eJournal of Phytopathology.\u003c/em\u003e https://doi.org/10.1111/jph.70084\u003c/li\u003e\n\u003cli\u003eNguyen, L. L. H., Khuu, D. T., \u0026amp; Nguyen, T. Q. (2023). Factors influencing adoption of precision agriculture. \u003cem\u003eEvaluation Review, 48\u003c/em\u003e(4). https://doi.org/10.1177/0193841X231200775\u003c/li\u003e\n\u003cli\u003eOdounfa, M. G. F., et al. (2025). Deep learning enables precision agriculture for chili pepper disease detection in Benin. \u003cem\u003eDiscover Artificial Intelligence, 5,\u003c/em\u003e 315.\u003c/li\u003e\n\u003cli\u003eOmotayo, A. O., Adediran, S. A., Omotoso, A. B., Olagunju, K. O., \u0026amp; Omotayo, O. P. (2025). Artificial intelligence in agriculture: Ethics, impact possibilities, and pathways for policy. \u003cem\u003eComputers and Electronics in Agriculture, 239,\u003c/em\u003e 110927. https://doi.org/10.1016/j.compag.2025.110927\u003c/li\u003e\n\u003cli\u003e\u0026Ouml;zoğul, G. (2025). Applications of artificial intelligence technologies in agriculture: Advantages, challenges, risks, prospects, and recommendations. \u003cem\u003eCogent Food \u0026amp; Agriculture, 11\u003c/em\u003e(1). https://doi.org/10.1080/23311932.2025.2568199\u003c/li\u003e\n\u003cli\u003ePadhiary, M., Saha, D., Kumar, R., Sethi, L. N., \u0026amp; Kumar, A. (2024). Enhancing precision agriculture: A comprehensive review of machine learning and AI vision applications in all-terrain vehicle for farm automation. \u003cem\u003eSmart Agricultural Technology, 8,\u003c/em\u003e 100483. https://doi.org/10.1016/j.atech.2024.100483\u003c/li\u003e\n\u003cli\u003eParganiha, V., \u0026amp; Verma, M. (2025). An efficient disease prediction in smart agriculture using advanced deep learning methods for improving crop productivity. \u003cem\u003eJournal of Phytopathology.\u003c/em\u003e https://doi.org/10.1111/jph.70160\u003c/li\u003e\n\u003cli\u003ePearson, S. (2025). Artificial intelligence in horticulture. \u003cem\u003eThe Journal of Horticultural Science and Biotechnology, 100\u003c/em\u003e(5), 564\u0026ndash;570. https://doi.org/10.1080/14620316.2025.2475401\u003c/li\u003e\n\u003cli\u003eRashid, A. B., Kausik, A. K., Khandoker, A., \u0026amp; Siddque, S. N. (2025). Integration of artificial intelligence and IoT with UAVs for precision agriculture. \u003cem\u003eHybrid Advances, 10,\u003c/em\u003e 100458. https://doi.org/10.1016/j.hybadv.2025.100458\u003c/li\u003e\n\u003cli\u003eRyan, M., Isakhanyan, G., \u0026amp; Tekinerdogan, B. (2023). An interdisciplinary approach to artificial intelligence in agriculture. \u003cem\u003eNJAS: Impact in Agricultural and Life Sciences, 95\u003c/em\u003e(1). https://doi.org/10.1080/27685241.2023.2168568\u003c/li\u003e\n\u003cli\u003eSaiz, E., Iqbal, F., Grant, J. H., Korir, L., Ullah, S., \u0026amp; Radu, A. (2026). Technological and socio-economic challenges in the development of sensors for precision agriculture. \u003cem\u003eAgriculture, 16\u003c/em\u003e(1), 89. https://doi.org/10.3390/agriculture16010089\u003c/li\u003e\n\u003cli\u003eSood, A., Bhardwaj, A. K., \u0026amp; Sharma, R. K. (2022). Towards sustainable agriculture: Key determinants of adopting artificial intelligence in agriculture. \u003cem\u003eJournal of Decision Systems, 33\u003c/em\u003e(4), 833\u0026ndash;877. https://doi.org/10.1080/12460125.2022.2154419\u003c/li\u003e\n\u003cli\u003eSu, K.-W., Wang, C.-H., Lin, O. Y.-H., Chen, H.-J., Wang, W.-L., \u0026amp; Chiu, P.-C. (2026). Human-centered design of an artificial intelligence-based diagnostic system for tomato diseases in smallholder agriculture. \u003cem\u003eJournal of Industrial and Production Engineering.\u003c/em\u003e https://doi.org/10.1080/21681015.2026.2613023\u003c/li\u003e\n\u003cli\u003eTran Cao Uy, Limnirankul, B., Kramol, P., Hoang Gia, H., \u0026amp; Nguyen Thi, D. T. (2024). Digital technology adoption among smallholder farmers in Vietnam: Implications for digital agricultural extension strategies. \u003cem\u003eJournal of International Development.\u003c/em\u003e https://doi.org/10.1002/jid.3904\u003c/li\u003e\n\u003cli\u003eUgwu, O. P.-C., Ogenyi, F. C., Alum, E. U., Eze, V. H. U., Basajja, M., \u0026amp; Ugwu, J. N. (2025). Implementing artificial intelligence and machine learning algorithms for optimized crop management: A systematic review on data-driven approaches to enhancing resource use and agricultural sustainability. \u003cem\u003eCogent Food \u0026amp; Agriculture, 11\u003c/em\u003e(1). https://doi.org/10.1080/23311932.2025.2569982\u003c/li\u003e\n\u003cli\u003eUpadhyay, A., Chandel, N. S., Singh, K. P., Chakraborty, S. K., Nandede, B. M., Kumar, M., Subeesh, A., Upendar, K., Salem, A., \u0026amp; Elbeltagi, A. (2025). Deep learning and computer vision in plant disease detection: A comprehensive review of techniques, models, and trends in precision agriculture. \u003cem\u003eArtificial Intelligence Review, 58,\u003c/em\u003e 92. https://doi.org/10.1007/s10462-024-11100-x\u003c/li\u003e\n\u003cli\u003eWu, P., \u0026amp; Zhong, Y. (2025). Artificial intelligence in sustainable agriculture: Towards a socio-technical roadmap. \u003cem\u003eSmart Agricultural Technology, 12,\u003c/em\u003e 101578. https://doi.org/10.1016/j.atech.2025.101578\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"discover-global-society","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"Learn more about [Discover Global Society](https://www.springer.com/journal/44282)","snPcode":"44282","submissionUrl":"https://submission.nature.com/new-submission/44282/3","title":"Discover Global Society","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Discover Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Artificial intelligence in agriculture, Precision agriculture, Smallholder farmers, Technology adoption, Sustainable agricultural productivity","lastPublishedDoi":"10.21203/rs.3.rs-8808017/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8808017/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThe application of artificial intelligence (AI) to precision agriculture has become a revolutionary tool for improving the productivity, sustainability, and resiliency of smallholder agricultural systems, especially in developing and rural settings. This paper will discuss the use of AI tools, including machine learning-based crop disease detection, yield prediction models, smart irrigation, and decision-support systems, in the context of smallholder farming based on a systematic review of 50 peer-reviewed articles that were published by reputable journals indexed in Elsevier/ScienceDirect, Taylor and Francis, Wiley, SAGE, and Springer Nature. The review also provides consistent evidence that AI applications have the potential to positively transform agricultural productivity by diagnosing diseases earlier, better using inputs, and managing farms based on data, as well as enhancing environmental sustainability. Nonetheless, the adoption by the smallholder farmers is still disproportionate and highly contextual. There are perceptions and intent to adopt AI-based technologies that are behavioral and socio-economic in nature and are strongly influenced by perceptions and attitudes to AI systems, trust, digital literacy, access, and cost of data infrastructure, and institutional support. The literature also points to the existence of severe obstacles like low levels of connectivity, skills gaps, disjointed extension services, ethical and data-governance issues, and the inaccessibility of high-tech solutions and smallholder realities. Meanwhile, the uptake and impact can be greatly improved with the help of the enabling factors, such as human-centered design, advisory and extension services, facilitating policies, and inclusive innovation ecosystems. This study contributes to the holistic insight into AI-enabled precision agriculture among smallholders by incorporating both technical performance evidence and socio-economic as well as policy viewpoints. It offers a conceptual basis and a research focus for future empirical investigations, stating that future AI solutions to optimize the advantages of digital agriculture must be context-aware, equitable, and farmer-focused to make sure that the advantages of digital agriculture are widely distributed across rural communities.\u003c/p\u003e","manuscriptTitle":"AI-Enabled Precision Agriculture for Smallholder Farmers","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-02-12 08:44:39","doi":"10.21203/rs.3.rs-8808017/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-04-17T04:26:10+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-03-26T21:13:01+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"220878908961096048825970547728166208391","date":"2026-03-18T14:00:57+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"327432813692208904502300838456159759116","date":"2026-03-14T18:06:00+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-03-11T16:16:38+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"117638822873301073674353593354159659324","date":"2026-03-01T14:35:38+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-02-27T14:32:35+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2026-02-27T13:58:37+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-02-09T14:27:16+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-02-09T14:26:31+00:00","index":"","fulltext":""},{"type":"submitted","content":"Discover Global Society","date":"2026-02-06T13:43:36+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"discover-global-society","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"Learn more about [Discover Global Society](https://www.springer.com/journal/44282)","snPcode":"44282","submissionUrl":"https://submission.nature.com/new-submission/44282/3","title":"Discover Global Society","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Discover Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"0279459f-3452-4660-93a5-e36bc0c3e3f7","owner":[],"postedDate":"February 12th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"in-revision","subjectAreas":[],"tags":[],"updatedAt":"2026-05-18T05:39:34+00:00","versionOfRecord":[],"versionCreatedAt":"2026-02-12 08:44:39","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8808017","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8808017","identity":"rs-8808017","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}