Are Professionals also influenced by AI to Make Impulse Buying? A Dual-System Cognitive Process of Sports-Specialised Customers

Are Professionals also influenced by AI to Make Impulse Buying? A Dual-System Cognitive Process of Sports-Specialised Customers | 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 Article Are Professionals also influenced by AI to Make Impulse Buying? A Dual-System Cognitive Process of Sports-Specialised Customers Qingguang Meng, Qiongfeng Feng, Wenhui Yue, Qingqing Zheng, Hongtian Huang This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9008587/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract In the context of human–AI collaborative consumption, AI chatbots have emerged as a critical tool for enhancing consumer conversion rates on e-commerce platforms. Their influence on impulsive buying behaviour among general consumers has garnered significant academic and industry attention. However, the extent to which AI-driven recommendation systems may influence expert judgment requires further empirical investigation. To investigate the impact of interactions between sports-specialised consumers and AI chatbots on impulsive buying behaviour with respect to athletic training products, this study integrates the S-O-R model with dual-systems theory and constructs a dual-path cognitive mechanism. It explores the mediating roles of flow experience and perceived risk, as well as the moderating effects of sports-specialised consumers, within the dual-path framework. The results indicate that AI service quality positively influences the impulsive buying tendencies of sports-specialised consumers through flow experience, that AI service quality negatively affects their impulsive buying tendencies through risk perception, and that customers with sports-related expertise serve as a moderating variable, attenuating the positive effect of flow experience induced by AI chatbots and amplifying the positive influence of perceived risk. This study elucidates the cognitive processing pathways of sports-specialised consumers in human–computer interactions and provides empirical evidence to assist e-commerce enterprises in optimising AI service design for targeted demographic segments. Business and commerce/Business and management Social science/Business and management Business and commerce/Information systems and information technology Biological sciences/Psychology Social science/Psychology Social science/Science technology and society AI chatbots Impulsive buying behaviour Flow experience Perceived risk Sports-specialised consumers Figures Figure 1 Figure 2 Figure 3 Figure 4 1. Introduction AI chatbots, leveraging 24/7 real-time responsiveness and personalised recommendation algorithms, have emerged as pivotal tools in e-commerce for enhancing sales conversion rates and optimising service efficiency(Song, Xing, Duan, Cohen, & Mou, 2022). The implementation of AI chatbots has not only significantly enhanced the consumer buying experience and stimulated impulse buying behaviour but also contributed to substantial revenue growth for e-commerce enterprises. However, the efficacy of AI chatbots in enhancing purchase intention among professional consumers in specialised sectors, such as sports-specialised consumers, still requires further empirical investigation. Sports training equipment is highly specialised, and consumers in the sustainable consumption lifecycle demonstrate more detailed demands for product functionalities, materials, and associated technical specifications(Spindler, Schunk, & Könecke, 2023). Particularly among the young professional consumer demographic, a propensity for instant gratification is coupled with consumption behaviours that prioritise experiential engagement and social connectivity(Coelho, Aniceto, Bairrada, & Silva, 2023). In this context, e-commerce enterprises should leverage AI interaction technologies to deliver expert, real-time informational support and personalised shopping recommendations, thereby enhancing the consumer’s retail experience. Existing research has demonstrated that external stimuli such as live-streaming interactions and augmented reality can effectively trigger impulse buying(Xiaoping Zhang & Zhang, 2024). AI chatbots also possess the ability to influence consumer decision-making, but their impact on purchasing decisions varies depending on the service level of the AI chatbot. Crucially, even when professionals make purchases influenced by AI chatbots, they often return products later because of mismatches with expectations, resulting in high after-sales costs and operational pressures for e-commerce enterprises(Shao, Cheng, Wan, & Yue, 2021). Therefore, exploring the underlying mechanisms through which AI chatbots influence impulse purchasing behaviour to increase corporate sales efficiency in human–machine collaborative consumption scenarios has become a hot topic of sustained theoretical interest(J.-C. Lee & Xiong, 2024). The literature has begun to focus on the positive effects of AI chatbot communication quality; however, it has overlooked the conflicting effects that may arise from different cognitive mechanisms. In theory, AI chatbots can accurately capture customers' intentions, preferences, and emotions during conversations(Trivedi, Kasilingam, Arora, & Soni, 2022), providing consumers with instant service(Song et al., 2022). Through their human-like, humorous style, AI chatbots increase consumer satisfaction with service during the shopping process, thereby increasing purchasing intention(Shin, Bunosso, & Levine, 2022; Tan, Li, Huang, & Liu, 2025). Moreover, AI chatbots’ precise personalised recommendation capabilities can push products tailored to consumer needs on the basis of browsing history and preference data, sparking consumer interest and driving impulse buying(S. Chen, Li, Liu, & Wang, 2023). These studies implicitly assume that consumers are passive recipients of AI-generated information, thus overlooking people’s capacity for deep processing of information and autonomous decision-making. With respect to the cognitive processing of professional consumers of sporting goods, dual-systems theory posits that the human mind comprises two distinct cognitive systems: System 1 operates as an unconscious, automatic mode of thinking, relying on experience, emotions, and associative memory to generate rapid responses, whereas System 2 functions as a rational, conscious mode of thinking that requires focused attention to execute cognitive processing and self-regulation(Milli, Lieder, & Griffiths, 2021). During the process of AI chatbots providing communication services, emotional stimuli activate System 1 to trigger a flow experience, immersing consumers in a pleasurable state. This state reduces the cognitive responses associated with rational decision-making(Sengoz, Cavusoglu, Kement, & Bayar, 2024), thereby driving consumers toward the intuitive impulse buying of sports products. Cognitive stimuli activate System 2, thereby triggering perceived risk. Consumers then carefully consider potential product issues, usage risks, and after-sales guarantees(I.-L. Wu, Chiu, & Chen, 2020; Xiaoping Zhang & Zhang, 2024), which may increase their perceived risk of the product and enable rational decision-making for the impulse buying of sporting goods. In this study, a research framework based on S-O-R theory and dual-system theory is constructed, and AI chatbot service quality is treated as an external stimulus (S). By activating consumers’ flow experiences and physiological responses to perceived risk (O), this stimulus ultimately influences consumers’ impulse buying behaviour toward sporting goods (R). This study reveals a dual-path mechanism through which the communication quality of AI chatbots influences impulse buying: Flow experiences catalyse impulse buying via emotional responses, whereas perceived risk inhibits it through rational deliberation. Together, these elements constitute the decision-making logic in human‒machine collaborative consumption scenarios. The research contributions of this paper are as follows: (1) It pioneers an exploration of the mechanism through which AI (machine) influences customers’ (human) online impulse buying. The study reveals that System 2 cognitive processing heightens perceived risk during consumer interactions with AI chatbots, thereby inhibiting impulse buying behaviour. These findings respond to the research call by Wu et al. (2021) regarding how rational information affects online impulse buying by expanding the theory of human–machine interaction(Y. Wu, Xin, Li, Yu, & Guo, 2021). (2) Comparing professional judgment with AI precision recommendations, this study pioneers the comparison of human expertise versus AI expertise in new consumption scenarios. Its examination of the dual-system cognitive process underlying impulse buying among sports professionals explains the unique decision-making of experts while responding to calls by Artem Redine et al. to explore consumer impulse buying in new domains(Redine, Deshpande, Jebarajakirthy, & Surachartkumtonkun, 2022). 2. Theoretical Foundations and Research Hypotheses Dual-systems theory describes distinct cognitive processes occurring in System 1 and System 2(Barrouillet, 2011). System 1 relies on intuition, consumes fewer cognitive resources, and enables rapid, instinctive responses to stimuli; System 2, in contrast, analyses based on rules, requiring greater cognitive investment and complex computations(Milli et al., 2021). When System 1 dominates, individuals tend to make intuitive decisions; conversely, when System 2 takes precedence, it drives deeper thinking and rational judgment. The stimulus–organism–response (S-O-R) model further elucidates how external stimuli ultimately trigger specific behaviours or responses by influencing an individual’s internal state(Jacoby, 2002). The core principle of this model emphasises that an individual’s response is not directly determined solely by external stimuli but rather results from the combined effects of environmental stimuli, individual cognition, and emotional changes(Dalvi-Esfahani et al., 2025). Flow experience refers to an individual’s holistic state of engagement during an activity, often accompanied by intense pleasure and diminished self-awareness, allowing complete immersion in the present moment(C.-C. Chen & Lin, 2018; Nakamura & Csikszentmihalyi, 2009). In a flow state, consumers’ cognitive processing shifts to the System 1-dominant mode outlined in dual-system theory. Consumers tend to process stimuli delivered by AI chatbots intuitively and automatically, ultimately responding to such stimuli with impulsive buying behaviour. In contrast, under perceived risk conditions, consumers’ cognitive patterns shift toward the System 2-dominant mode. When confronted with information or recommendations from AI chatbots, consumers tend to evaluate potential risks through rational thinking(Cabeza-Ramírez, Sánchez-Cañizares, Santos-Roldán, & Fuentes-García, 2022), ultimately employing risk-averse decision-making logic to reduce impulsive buying behaviour. Additionally, when consumers are confronted with external stimuli from AI chatbots, their identity characteristics may also influence their cognitive processing pathways. Consumers possessing specialised knowledge can leverage their understanding of products and domains to more accurately identify potential consequences that AI chatbot stimuli may trigger. Therefore, this study integrates dual-system theory with the S-O-R model to examine the influence mechanisms of flow experiences and perceived risk on impulse buying behaviour toward sports products while further exploring the moderating role of sports-specialised consumers’ identity. 2.1 Communication Quality of AI Chatbots and Impulse Buying: The Mediating Role of Flow Experience The S-O-R model indicates that external environmental stimuli act upon an individual’s psychological and cognitive processes, ultimately triggering specific behavioural responses(Dalvi-Esfahani et al., 2025; Feng, Al Mamun, Masukujjaman, & Yang, 2023). In e-commerce shopping scenarios, AI chatbots can be regarded as external stimulus variables. Leveraging communication advantages such as instant responses and human-like interactions can reduce consumers’ cognitive processing(Song et al., 2022). On the one hand, AI chatbots offer round-the-clock automated communication that can respond to consumer inquiries at any time, thereby effectively reducing waiting periods. This process not only diminishes users’ perception of time passing but also decreases the cognitive processing involved in actively searching for and analysing information, making it easier for consumers to enter a flow experience state dominated by intuition. Research has indicated that the immediacy and convenience of interacting with AI bots are key technological features that drive consumer immersion in online retail scenarios(Z. Cheng, Fan, Shao, Jia, & Zhang, 2024). In this low-cognitive-load communication environment, consumers are more inclined to rely on intuitive cognition to process information, thereby increasing the likelihood of experiencing flow states. On the other hand, with the advancement of machine learning technology, AI chatbots can construct accurate user profiles based on customer retention data, precisely identify user needs, and provide appropriate responses(Rizomyliotis, Kastanakis, Giovanis, Konstantoulaki, & Kostopoulos, 2022), thereby providing a more tailored interactive experience for consumers. This anthropomorphic interaction not only enhances the approachability of the service but also allows users to experience emotional support and pleasure during communication(Alalwan, Algharabat, Abu El Samen, Albanna, & Al-Okaily, 2025), thereby strengthening their flow experience throughout the interaction process. Moreover, the personalised recommendations provided by AI chatbots can further enhance consumer engagement, keeping them highly focused throughout the information-gathering and communication process(S. Chen et al., 2023). These high-quality communication features of AI chatbots reduce cognitive resistance among consumers, making them more likely to ignore external distractions and fully immerse themselves in interactions with the AI chatbot, thereby triggering a flow experience. On this basis, the following hypothesis is proposed in this study. H1 : The communication quality of AI chatbots has a positive effect on flow experiences. Dual-system theory posits that individuals possess two distinct cognitive processing systems. System 1 represents an intuitive cognitive processing mode characterised by low cognitive load and reliance on emotion-driven mechanisms(Barrouillet, 2011; Milli et al., 2021). The core characteristic of flow experience lies in the individual’s complete immersion in the present activity and diminished self-awareness(Nakamura & Csikszentmihalyi, 2009), a psychological state that aligns with System 1’s cognitive processing mode. Specifically, consumers in a flow state experience heightened pleasure, and immersion in this emotional state weakens their capacity for behavioural reflection, thus making them more inclined to immediate, emotional responses to external stimuli. Within this cognitive context, System 1 processing dominates, automating consumers’ cognitive processing of information—such as instant communication services and promotional offers—provided by AI chatbots. This state reduces consumers’ rational evaluation of information, making them more susceptible to unplanned impulse buying. Furthermore, the pleasure and enjoyment accompanying flow experiences trigger an affective cognitive processing mode(Sengoz et al., 2024; Xiaoping Zhang & Zhang, 2024). In this state, consumers unconsciously project these emotions onto their current consumption context and the products they encounter. This emotion-driven cognitive processing significantly reduces consumers’ perception of potential product risks(Xiaoping Zhang & Zhang, 2024), weakens their rational risk-averse tendencies, and further catalyses impulsive buying decisions. Notably, during interactions with AI chatbots, multiple external stimuli—such as automatic coupon pushes, limited-time promotion pop-ups, and human-like communication tones—mutually reinforce consumers’ flow experiences. These stimuli reduce cognitive processing, deepen the flow experience, and prompt consumers to base their judgments of product value on immediate emotional responses. Ultimately, this influence leads to impulsive buying behaviour. On this basis, the following hypothesis is proposed in this study. H2 : Flow experiences have a positive effect on impulse buying. In combination, H1 and H2 suggest that the communication quality of AI chatbots enhances consumers’ flow experience, which in turn further drives their impulse buying behaviour. On this basis, the following hypothesis is proposed in this study. H3: Flow experience mediates the relationship between communication quality in AI chatbots and impulse buying. 2.2 Communication Quality of AI Chatbots and Impulse Buying: The Mediating Role of Perceived Risk According to dual-system theory, when consumers encounter complex or potentially threatening external stimuli, their cognitive processing mode shifts to System 2, which requires rational analysis(Barrouillet, 2011; Milli et al., 2021). During professional consultations about sporting goods, the limitations of AI chatbot responses and their collection of private information drive consumers into the rational cognitive processing mode of System 2, thereby amplifying their perceived risks associated with sporting goods. Specifically, consumers place high demands on safety and functionality when they purchase sporting goods(Chiu, Kim, & Won, 2018). Although AI chatbots can provide instant communication services(Z. Cheng et al., 2024), their performance in e-commerce contexts often fails to meet consumers’ specialised needs and may even exacerbate negative perceptions(Y. Zhang et al., 2025). A seamless conversational experience tends to increase consumers’ expectations regarding the communication capabilities of AI chatbots. When these AI chatbots fail to address complex inquiries specific to particular sports scenarios, this cognitive dissonance can erode consumer trust in the AI chatbot. Consequently, consumers may rationally evaluate the accuracy and completeness of the information provided by the AI chatbot, which ultimately heightens their perception of the risk associated with the product. Additionally, while AI chatbots can provide personalized recommendations 10 , they still collect users’ personal information—such as exercise preferences, purchase history, and physical data—during communication to achieve more precise conversations and recommend suitable sports products. Research has confirmed that the collection of private information triggers users’ risk perception(Y. Cheng & Jiang, 2020). Once consumers detect in-depth collection and analysis of such personal data during interactions, this recognition activates the rational cognitive processing of System 2, prompting them to weigh the convenience of personalised services against the potential risks of data breaches. This cognitive process amplifies consumers’ trust concerns about AI chatbot services, thereby increasing perceived risk(N. Chen & Yang, 2023; Chiu, Cho, & Chua, 2023). On this basis, the following hypothesis is proposed in this study. H4: The communication quality of AI chatbots has a positive effect on perceived risk. Perceived risk refers to an individual’s subjective assessment of the uncertainties associated with a specific product and the potential negative consequences they may trigger(Dholakia, 2001; Xiaoxue Zhang & Yu, 2020). System 2 cognitive processing further translates this assessment into concrete risk-avoidance strategies. Impulse buying is essentially a hedonistic, immediate behavioural response directly triggered by external stimuli(I.-L. Wu et al., 2020), while high levels of perceived risk inhibit this behaviour through various mechanisms. On the one hand, perceived risk prompts consumers to reflect on the suitability of the product for their needs and their own purchasing desires(Grünzner, Richter, White, & Pahl, 2025), which compels them to continuously allocate limited cognitive resources toward risk assessment and thereby dampens immediate buying impulses(Cabeza-Ramírez et al., 2022). On the other hand, high levels of perceived risk drive consumers to rationally weigh the actual utility of the product against potential return costs, thus effectively curbing impulsive buying behaviour. The use of sporting goods is directly linked to consumers’ physical health and safety. The safety and functional characteristics of sporting goods mean that consumers are more risk-sensitive toward these products than toward ordinary products. Therefore, even when enticed by marketing stimuli such as time-limited discounts and pop-up ads pushed by AI chatbots in e-commerce contexts(Z. Cheng et al., 2024), consumers’ concerns about information asymmetry regarding products, worries about after-sales guarantees(Chiu et al., 2023; Y. Zhang et al., 2025), and doubts about the reliability of AI chatbot services will still be processed rationally by System 2 into risk perceptions. This perceived risk ultimately negatively influences impulsive buying behaviour. On this basis, the following hypothesis is proposed in this study. H5: Perceived risk has a negative effect on impulse buying. In combination, H4 and H5 suggest that the communication quality of AI chatbots increases consumers’ perceived risk, which in turn further weakens their impulse buying behaviour. Therefore, this study proposes the following hypothesis. H6: Perceived risk mediates the relationship between AI chatbot communication quality and impulse buying. 2.3 The Moderating Role of Sports-Specialised Consumers Flow experience is a psychological state in which an individual’s attention is absorbed and immersed in an activity(Nakamura & Csikszentmihalyi, 2009), dominated by intuitive System 1 cognitive processing. When customers with a sports-specialised background interact with AI chatbots, they do not passively receive information but instead process the messages conveyed by the AI chatbot. This process keeps System 2 continuously engaged, prompting rational analysis of the external stimuli generated by the AI chatbot(Milli et al., 2021). Research has confirmed that professional consumers possess more complex knowledge structures in their memory, enabling them to analyse and process product information more deeply than novice consumers do(Chinchanachokchai, Thontirawong, & Chinchanachokchai, 2021). This understanding enables sports-specialised customers to evaluate product information or recommendations provided by AI chatbots on the basis of their own expertise. They are not swayed by the superficial fluency of AI conversations or marketing rhetoric, nor do they blindly accept the chatbot’s suggestions(Buechner, Stadler Blank, Escoe, & Blaney, 2024; Kim, Kim, & Lee, 2025). This rational approach diminishes the ability of the AI chatbot to enhance flow experiences through communication quality. Furthermore, flow experiences encompass emotional elements, such as pleasure and enjoyment(Sengoz et al., 2024). AI chatbots often employ anthropomorphic expressions and emotional interactions to enhance engagement(Song et al., 2022), thereby stimulating positive user emotions and immersion. In sports product purchasing decisions, sports-specialised customers place greater emphasis on the reliability of their own assessments of product quality and performance(Uhm, Kim, Do, & Lee, 2022). Consequently, they expect AI chatbots to deliver precise, professional, and practical information to help them make informed product decisions. If an AI chatbot’s communication approach emphasises emotional interaction over substantive professional support, it will not only fail to attract professionals but may even be perceived as disruptive or unprofessional. Consequently, in sports product e-commerce purchasing scenarios, the formation of flow experiences during interactions between sports-specialised customers and AI chatbots will be inhibited. On this basis, the following hypothesis is proposed in this study. H7: Sports-specialised customers moderate the relationship between AI chatbot communication quality and flow experience, with sports-specialised customers weakening the positive influence of AI chatbot communication quality on the flow experience. According to dual-system theory, sports-specialised customers’ knowledge base leads them to rely more heavily on the rational analytical pathway of System 2 when evaluating the communication quality of AI chatbots(Barrouillet, 2011; Milli et al., 2021). This cognitive processing model enables it to keenly identify implicit information biases in AI conversations, thereby enhancing the recognition of perceived risks. Even if AI chatbots appear to function flawlessly during communication, they can still trigger doubts among sports-specialised customers about the authenticity of information and the suitability of product recommendations. Such perceptions lead customers to reject AI-generated suggestions, thereby amplifying perceived risk. Compared with general consumers, sports-specialised customers have higher expectations for the quality and performance of athletic products(Spindler et al., 2023). Their evaluation of AI communication quality centres on the accuracy of the content conveyed rather than solely focusing on the interactive experience. Therefore, when AI chatbots attempt to enhance service quality through personalised recommendations, sports-specialised customers will verify the validity of these suggestions on the basis of their own expertise(Chinchanachokchai et al., 2021). If they find that the recommended products deviate from their actual athletic needs, they will not only reject the information provided by the AI chatbot but also perceive this proactive service as misleading. This evaluation further amplifies their perception of risk with respect to product functionality. Additionally, in online shopping scenarios, consumers cannot physically touch or experience products. AI chatbots primarily convey information through text, making it difficult to enhance the virtual experience through immersive design and multisensory displays such as AR technology(Uhm et al., 2022). This limitation makes it challenging for sports-specialised customers—who pay close attention to product details—to verify actual functionality on the basis solely of textual descriptions. As a result, they are highly prone to distrust information provided by AI chatbots, which ultimately exacerbates the perception of risk. On this basis, the following hypothesis is proposed in this study. H8: Sports-specialised customers moderate the relationship between AI chatbot communication quality and perceived risk. Sports-specialised customers enhance the positive influence of AI chatbot communication quality on perceived risk. Based on the above assumptions, the conceptual model of this study is shown in Fig 1. Fig 1. Research model 3. Data and Methods 3.1 Sample and Data Collection To ensure that participants had a thorough understanding of the survey content and to guarantee the reliability and validity of the data, this study targeted consumers with online shopping experience. Specifically, a portion of the sample was restricted to individuals with professional knowledge and skills in the sports field. To minimise common method bias during questionnaire distribution, this study employed a two-stage time-lagged design for data collection. The first phase involved the distribution of questionnaires via the Tencent survey platform, primarily measuring respondents’ demographic characteristics, such as gender, age, years of online shopping experience, and online shopping frequency. Additionally, it assessed the independent variable of communication quality with AI chatbots and the moderating variable of sports- specialised customers. A total of 527 questionnaires were distributed during this phase. After invalid samples were excluded, 432 valid questionnaires were ultimately collected. Two weeks after the first round of surveys concluded, we invited valid participants who had completed Phase One to participate in Phase Two. In this phase, the questionnaire included core variable measurement items such as flow experience, perceived risk, and impulse buying behaviour. In the second phase, a total of 432 questionnaires were distributed. After further screening for invalid responses, 304 valid questionnaires were ultimately obtained. To ensure the precise matching of data across both phases while strictly protecting participant privacy, we requested that participants provide the last four digits of their mobile phone number as an anonymous identifier in both phases. Full contact details were immediately deleted after data cleaning and matching to guarantee that no traceable personal identification remained. Furthermore, all the questionnaires prominently displayed the research-informed consent form on the first page, detailing the study objectives, data usage, and confidentiality measures. Participants only proceeded to the formal questionnaire after providing their consent. Among the valid samples, males accounted for 56.91% and females for 43.09%, indicating a male-dominated sample with a relatively balanced gender ratio. The participants were predominantly young and middle-aged adults, with the greatest proportion (32.57%) aged 31–35, followed by those aged 26–30 (26.97%). Those aged 21–25 accounted for 19.08%, while those aged 36 and above accounted for 21.38%. This age structure indicates that the core sample consisted of mature users with strong purchasing power. The vast majority of participants had more than one year of online shopping experience: 38.16% had 1–3 years, 28.29% had 3–5 years, and 22.70% had more than 5 years. This indicates the sample’s overall familiarity with online shopping processes and environments, which enhanced their comprehension of survey content and the quality of their responses. Nearly three-quarters of the respondents shopped online 3 or more times per month, with 3–5 times being the most common at 37.17%, followed by 6–10 times at 26.32%, and more than 10 times at 10.20%. This indicates that sample users generally exhibited active online consumption habits, meeting the study’s screening requirement for “having online shopping experience.” 3.2 Measurement Items All the variables in this research model were selected from established scales published in authoritative academic journals. During the translation process, we implemented a rigorous back-translation procedure conducted collaboratively by bilingual experts. Through this process, items exhibiting semantic discrepancies were discussed and revised, ultimately resulting in the formal survey questionnaire used in this study. 3.2.1 Dependent Variable The measurement of impulse buying was primarily based on the research by Beatty and Elizabeth (1998) and incorporated measurement items from Xin et al. ( 2025 )(Beatty & Elizabeth Ferrell, 1998 ; Xin, Jian, Liu, & Bao, 2025 ). The instrument comprises three items, with a typical example being “I am someone who buys products that weren’t originally planned.” 3.2.2 Independent Variable The measurement of AI chatbot communication quality employed a scale developed by Chung et al. ( 2020 ), drawing upon the research of Lee and Park ( 2022 )(Chung, Ko, Joung, & Kim, 2020 ; M. Lee & Park, 2022 ). It comprises 12 items, with a typical example being “When shopping online, my communication with the AI chatbot is timely.” 3.2.3 Mediating Variables The measurement of flow experience employed the scale developed by Chang and Zhu ( 2012 )(Chang & Zhu, 2012 ), drawing upon research by Chen and Lin ( 2018 ) et al.(C.-C. Chen & Lin, 2018 ). It comprises four items, with a typical example being “When shopping online, I feel time passes quickly when communicating with the AI chatbot.” Perceived risk was measured on the basis of Dholakia’s ( 2001 ) study(Dholakia, 2001 ), incorporating items from Cabeza-Ramírez et al. ( 2022 )(Cabeza-Ramírez et al., 2022 ). It comprises three items, with a representative example being: “It’s risky to buy products recommended/promoted by AI chatbots.” 3.2.4 Regulating Variables The moderating variable in this study was sports-specialised customers. To avoid suggestive bias from direct questioning in the questionnaire design, we measured this variable with a background screening item: “What is your professional field?” The options were as follows: “Sports field: holding nationally certified sports qualifications, possessing a higher education background in sports, or currently engaged in sports-related professional work,” “Science, Engineering, Agriculture, and Medicine: e.g., computer science, engineering, medicine, biology,” “Humanities and Social Sciences: e.g., literature, history, economics, law, arts,” “Other.” In subsequent data analysis, respondents selecting the sports field were coded as 1, indicating a professional sports background. All other respondents were coded as 0 and served as the control group. This approach concealed the purpose of the study while effectively distinguishing key moderating variables. 3.2.5 Control Variables This study controlled for gender, age, years of online shopping experience, and online shopping frequency. Except for the control variables and moderating variable, other variables were measured on a five-point Likert scale, where 1 denotes strongly disagree and 5 denotes strongly agree. 4. Results 4.1 Common Method Variance and Collinearity Test This study employed the Harman single-factor test to examine common method bias. The results revealed that the total variance explained by the sample data was 65.166%. The variance contribution rate of the first principal component obtained without rotation was 36.427%, which was below the 40% critical threshold. This result suggests that the sample did not exhibit severe common method bias. Collinearity testing was performed on the sample data using SPSS. As shown in Table 3, the highest VIF (1.384) among the variables was less than 5, indicating that no severe multicollinearity issues existed in the sample data. 4.2 Reliability and Validity Analysis In this study, Cronbach’s α coefficient was used to assess the reliability of the measurement scales for each variable. The results revealed that the Cronbach’s α values for AI chatbot communication quality, flow experience, perceived risk, and impulse buying behaviour were 0.940, 0.837, 0.814, and 0.773, respectively, with all the measurement items meeting the standard. Amos software was used to perform confirmatory factor analysis, with χ²/df, SRMR, RMSEA, NFI, TLI, and CFI as the core fit indices. The detailed results are presented in Table 1. Model comparisons revealed that as the number of factors decreased, all fit indices deteriorated. The four-factor model demonstrated optimal fit (χ²/df = 1.134, SRMR = 0.036, RMSEA = 0.021, NFI = 0.939, TLI = 0.991, CFI = 0.992). As indicated by Table 2, the square root of the average absolute value (AVE) consistently exceeds the absolute value of the Pearson correlation coefficients between variables, confirming the discriminant validity among these variables. Table 1. Results of confirmatory factor analysis (N=304) Model χ²/df SRMR RMSEA NFI TLI CFI Four-factor model （AIC, FE, PR, IB） 1.134 0.036 0.021 0.939 0.991 0.992 Three-factor model （AIC, FE+PR, IB） 2.847 0.094 0.078 0.845 0.880 0.893 Two-factor model （AIC, FE+PR+IB） 3.809 0.110 0.096 0.790 0.817 0.835 One-factor model （AIC+FE+PR+IB） 5.847 0.128 0.126 0.676 0.684 0.714 Notes: AIC=AI chatbot communication quality; FE=flow experience; PR=perceived risk; IB=impulse buying 4.3 Descriptive Statistics and Correlation Analysis To ensure that the applicability conditions for subsequent parameter tests are met, SPSS statistical software was used in this study to calculate the skewness and kurtosis coefficients of each measurement item, thereby examining the normal distribution characteristics of the sample data. The results revealed that the absolute values of skewness for all the measurement items in this study did not exceed the critical threshold of 3 and that the absolute values of the kurtosis coefficients were all less than 8. On this basis, the sample data can be deemed to meet the requirements for an approximate normal distribution. Furthermore, the VIF values for all the variables were less than 2, indicating no severe multicollinearity issues. The descriptive statistics and correlation analysis results are presented in Table 2. As shown in Table 2, all correlation coefficients fall within reasonable ranges. Specifically, AI chatbot communication quality was significantly positively correlated with flow experience (r = 0.270, p < 0.001) and with perceived risk (r = 0.275, p < 0.001). Flow experience was significantly positively correlated with impulse buying behaviour (r = 0.327, p < 0.001), whereas perceived risk was significantly negatively correlated with impulse buying behaviour (r = -0.219, p < 0.001). These findings provide preliminary validation of the research hypotheses. Table 2. Descriptive statistics and Pearson correlation analysis (N=304) 1 2 3 4 5 6 7 8 9 1. Gender 1.000 2. Age -0.178** 1.000 3. Years of Online Shopping 0.014 0.316*** 1.000 4. Online Shopping Frequency -0.019 0.077 -0.040 1.000 5. Sports-Specialised Customers -0.053 -0.012 -0.013 -0.057 1.000 6.AI Chatbot Communication Quality -0.109 0.003 -0.057 -0.030 0.292*** (0.758) 7. Flow Experience 0.109 -0.029 -0.059 -0.046 0.326*** 0.270*** (0.752) 8. Perceived Risk -0.055 -0.002 0.051 -0.021 -0.127* 0.275*** 0.195*** (0.772) 9. Impulse Buying 0.070 0.021 -0.021 -0.072 0.245*** 0.252*** 0.327*** -0.219*** (0.731) Mean 1.398 2.645 2.628 2.204 0.470 2.662 2.696 2.618 3.582 SD 0.490 1.171 0.953 0.946 0.500 1.039 1.160 1.200 1.019 VIF 1.087 1.169 1.140 1.020 1.288 1.355 1.384 1.368 1.346 Notes: * p<0.05, ** p<0.01, *** p<0.001; The numbers in diagonal brackets represent the square root of the AVE value. 4.4 Hypothesis Testing 4.4.1 Direct Effect Testing In this study, SPSS was employed to examine the mechanism through which AI chatbot communication quality influences impulsive buying behaviour via two distinct pathways. As shown by the regression analysis results in Table 4, in Model 1, AI chatbot communication quality significantly and positively affected flow experience (β = 0.281, p < 0.001), validating H1. In Model 8, flow experience significantly and positively influenced impulsive buying behaviour (β = 0.267, p < 0.001), validating H2. In Model 4, AI chatbot communication quality significantly and positively influenced perceived risk (β = 0.276, p < 0.001), validating H4. In Model 9, perceived risk exerted a significant negative effect on impulsive buying behaviour (β = -0.310, p < 0.001), validating H5. 4.4.2 Mediation Effect Test This study employed hierarchical regression analysis to separately examine the mediating effects of flow experience and perceived risk. As shown in Table 4, after flow experience was incorporated as a mediating variable in Model 8, AI chatbot communication quality still exerted a significant positive influence on impulse buying behaviour (β = 0.185, p < 0.001). However, compared with Model 7, the regression coefficient for AI chatbot communication quality decreased from 0.260 to 0.185. These findings indicate that flow experience mediated the relationship between AI chatbot communication quality and impulsive buying behaviour, validating Hypothesis H3. After perceived risk was incorporated as a mediating variable in Model 9, AI chatbot communication quality still exerted a significant positive effect on impulsive buying behaviour (β = 0.346, p < 0.001), with its regression coefficient increasing from 0.260 in Model 7 to 0.346. Moreover, perceived risk had a significant negative effect on impulsive buying behaviour (β = -0.310, p < 0.001). These findings indicate that perceived risk moderated the true relationship between AI chatbot communication quality and impulsive buying behaviour, validating Hypothesis H6. The bootstrap test results for the mediating effect (based on 5,000 samples) further revealed the underlying mechanism through which AI chatbot communication quality influences impulse buying. As shown in Table 3, the total effect of AI chatbot communication quality on impulse buying was 0.255, whereas its indirect effect via flow experience was 0.074. The 95% confidence interval did not contain zero, indicating that flow experience mediated the relationship between AI chatbot communication quality and impulse buying, thus validating H3. The indirect effect of AI chatbot communication quality on impulse buying via perceived risk was -0.084, with the 95% confidence interval not containing zero. These findings indicate that perceived risk mediated the relationship between AI chatbot communication quality and impulse buying, further validating H6. Table 3. Mediating effect test of the bootstrap method (N=304) Path Effect SE 95%CI LLCI ULCI Total effect 0.255 0.055 0.147 0.363 AI chatbot communication quality→ Flow experience→ Sports-specialized consumers Indirect effect 0.074 0.023 0.036 0.124 Direct effect 0.182 0.055 0.073 0.290 AI chatbot communication quality→ Perceived risk→ Sports-specialized consumers Indirect effect -0.084 0.024 -0.135 -0.044 Direct effect 0.339 0.054 0.232 0.446 In this study, Amos software was used to analyse the structural equation model. The overall model fit indices were as follows: χ²/df = 1.252, RMSEA = 0.029, GFI = 0.931, TLI = 0.984, CFI = 0.985, with all metrics meeting the ideal standards. The standardised path coefficients of the structural equation model and the significance levels of the relationships between the variables are presented in Fig 2. The results fully align with the hypotheses proposed in this study, further validating the relationships among AI chatbot communication quality, flow experience, perceived risk, and impulse buying. Fig 2. Standardised path coefficients in a structural equation model 4.4.3 Moderation Effect Test The moderator variable “sports-specialised customers” in this study involved categorical data. Therefore, before conducting stratified regression analysis, we first performed dummy variable coding. Simultaneously, the independent variable “AI chatbot communication quality” underwent centring. On the basis of the processed independent and moderator variables, we subsequently constructed their interaction term. As shown in Table 4, in Model 3, AI chatbot communication quality significantly positively influenced flow experience (β = 0.317, p < 0.001), whereas the moderating effect of sports-specialised customers was significantly negative (β = -0.163, p < 0.05). This finding indicates that the presence of sports-specialised customers weakened the positive effect of AI chatbot communication quality on flow experience, validating H7. In Model 6, AI chatbot communication quality significantly and positively influenced perceived risk (β = 0.199, p < 0.05). Concurrently, the moderating effect of sports-specialised customers was significantly positive (β = 0.201, p < 0.01), indicating that the presence of sports-specialised customers amplified the positive effect of AI chatbot communication quality on perceived risk. Thus, H8 is validated. The moderating effect of sports-specialised customers is illustrated in Fig 3 and 4. Fig 3 . The moderating effect of sports-specialised customers on AI chatbot communication quality and flow experience Fig 4 . The moderating effect of sports-specialised customers on AI chatbot communication quality and perceived risk Table 4 . Regression Analysis Results (N=304) Flow Experience Perceived Risk Impulse Buying Model1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Model 9 Gender 0.142* 0.149** 0.143** -0.032 -0.038 -0.031 0.106 0.068 0.097 Age 0.015 0.019 0.026 -0.032 -0.036 -0.044 0.053 0.049 0.043 Years of Online Shopping -0.052 -0.054 -0.048 0.077 0.079 0.072 -0.027 -0.013 -0.003 Online Shopping Frequency -0.038 -0.025 -0.032 -0.008 -0.019 -0.011 -0.068 -0.057 -0.070 AI Chatbot Communication Quality 0.281*** 0.202*** 0.317*** 0.276*** 0.342*** 0.199* 0.260*** 0.185*** 0.346*** Flow Experience 0.267*** Perceived Risk -0.310*** Non-Sports Specialised Consumers (Reference Group) 0.273*** Sports-Specialized Consumers 0.276*** -0.229*** -0.233*** AI Chatbot Communication Quality × Sports-Specialised Consumers -0.163* 0.201** R 2 0.096 0.164 0.176 0.082 0.130 0.149 0.080 0.144 0.168 Adjusted R 2 0.080 0.147 0.157 0.066 0.112 0.129 0.064 0.127 0.151 F 6.304*** 9.678*** 9.055*** 5.296*** 7.365*** 7.407*** 5.159*** 8.345*** 9.970*** Notes: *p < 0.05, **p < 0.01, ***p < 0.001 5. Discussion This study integrates the S-O-R framework with dual-process decision theory to reveal a dual-path mechanism through which AI chatbot communication quality influences impulse buying among sports-specialised customers. The research findings indicate that the communication quality of AI chatbots positively influenced impulse buying through flow experiences (H3) but negatively influenced impulse buying through perceived risk (H6). Specifically, the communication quality of AI chatbots significantly enhanced the flow experience (H1), which is consistent with the findings of Alalwan et al. ( 2025 ) (Alalwan et al., 2025 ). Flow experiences are characterised by immediate immersion and diminished self-awareness(Nakamura & Csikszentmihalyi, 2009 ). When stimulated by the instant responses of AI chatbots, consumers experience significantly reduced waiting costs, making them more susceptible to being drawn into conversations and engaging with promotional discounts, product recommendations, and other information pushed by AI chatbots(Z. Cheng et al., 2024 ). This immersive state weakens consumers’ rational reflection ability, ultimately promoting impulse buying (H2). Moreover, the communication quality of AI chatbots positively influences perceived risk (H4). Perceived risk stems from an individual’s subjective assessment of potential losses(Dholakia, 2001 ; Xiaoxue Zhang & Yu, 2020 ), and its formation is closely linked to System 2 rational cognitive processing. Although AI chatbots offer advantages such as instant responses and personalised service(Song et al., 2022 ), they may also trigger consumers’ rational decision-making because of issues such as information specificity and demand matching. For sports-specialised customers, their specialised knowledge reserves trigger System 2 rational processing to scrutinise AI-generated information. Upon detecting discrepancies, they intensify perceived risk, which thereby suppresses impulse buying (H5). This finding aligns with Wu et al.’s ( 2020 ) finding that perceived risk negatively affects online impulse buying(I.-L. Wu et al., 2020 ). This study further introduced sports-specialised customers as a moderating variable. The results indicated that sports-specialised customers both weakened the positive influence of AI chatbot communication quality on flow experience (H7) and strengthened its positive influence on perceived risk (H8). Moderation effect analysis revealed that the moderating effect of perceived risk among sports-specialised customers (β = 0.201, p < 0.01) was stronger than that of flow experience (β = -0.163, p < 0.05). This difference stems from the fact that sports-specialised customers prioritise product performance and safety in their buying decisions(Chiu et al., 2018 ), with this objective taking precedence over the emotional pleasure elicited by flow experience. Consequently, when they interact with AI chatbots, sports-specialised customers are more inclined to allocate their expertise to risk identification. This goal-oriented cognitive resource allocation results in a more pronounced moderating effect along the perceived risk pathway, while the influence is relatively weaker along the flow experience pathway, which relies on intuition and emotion. 5.1 Theoretical Contributions This study makes two key theoretical contributions: (1) It pioneers an exploration of the mechanism through which AI (machine) influences customers’ (human) online impulse buying, thereby expanding human‒computer interaction theory. On the basis of dual-system theory, this research reveals that System 2 cognitive processing heightens perceived risk during consumer interactions with AI chatbots, subsequently inhibiting impulse buying behaviour. This finding provides a novel perspective for understanding consumers’ dual cognitive processing mechanisms in human‒computer interaction scenarios. Existing research has predominantly focused on the positive experiences derived from the communication characteristics of AI chatbots while paying insufficient attention to the potential cognitive conflicts they may induce in consumers(Alalwan et al., 2025 ; Hao & Li, 2025 ). This study, which is grounded in dual-system theory, reveals the cognitive pathway through which AI chatbot communication quality influences purchase decisions. It finds that when interacting with AI chatbots, consumers simultaneously trigger System 1 intuitive processing because of communication fluency, leading to flow experiences, while also activating System 2 rational processing to evaluate the accuracy of AI-provided information, thereby heightening perceived risk. These findings corroborate prior research indicating that individuals are more receptive to AI recommendations in low-risk decision-making tasks(Longoni, Bonezzi, & Morewedge, 2019 ). However, in scenarios requiring expert knowledge assessment, AI recommendations may paradoxically diminish trust in AI judgments(Buechner et al., 2024 ), triggering scepticism about the information source. These findings not only clarify the rational decision-making pathway through which AI chatbot communication services influence consumer purchasing behaviour but also respond to scholars’ calls for further research on how rational information affects impulse buying(Y. Wu et al., 2021 ). (2) By comparing professional judgment with AI precision recommendations, this study pioneers the comparison of human expertise versus AI expertise in new consumption scenarios. Examining the dual-system cognitive process of impulse buying among sports-specialised customers, this study explains the unique decision-making of professionals. Focusing on specific consumption scenarios and groups, it expands the boundaries of consumer behaviour research. Previous studies have often targeted general consumers and lacked targeted exploration of specific groups. In this study, the sports e-commerce context forms its entry point, and sports-specialised customers are treated as a moderating variable to explore differences in cognition and behaviour during interactions with AI chatbots. The findings indicate that the moderating effect of sports-specialised customers is significant across both pathways. On the one hand, the high cognitive standards formed by sports-specialised customers on the basis of their expertise diminish their perception of AI interaction fluency, thereby weakening the positive impact of AI chatbot communication quality on flow experiences. On the other hand, their expertise enhances their rational thinking ability, enabling them to evaluate AI-provided information and thereby amplifying the positive influence of communication quality on perceived risk. This result stems from the decision-making characteristics of expert consumers, who can focus precisely on useful information during product evaluation to form rational decisions(Redine et al., 2022 ). Consequently, they can scrutinise product information and assess potential risks when they interact with AI chatbots. This study responds to scholars’ calls for further exploration of how consumer expertise influences AI product recommendation effectiveness and aligns with Artem Redine et al.’s ( 2022 ) proposal to extend impulse-buying research into new contexts(Chinchanachokchai et al., 2021 ; Redine et al., 2022 ). By comparing expert judgment with AI precision recommendations, this study provides a model for applying relevant theories to specific populations and specialised domains, thereby enhancing theoretical explanatory power and universality. 5.2 Practical Implications Relevant departments and industry associations should collaborate to establish information service standards for AI chatbots in specialised fields such as sporting goods. These standards should specify the information content that AI chatbots must disclose during product recommendations, thereby driving technological development aligned with industry needs. Concurrently, interdisciplinary communication platforms should be established to mitigate information biases in AI chatbot services within specialised domains, thereby fostering a regulated human‒machine interaction environment for consumer activities. Businesses can implement differentiated AI chatbot service strategies for distinct customer segments to meet their varied communication needs. For customers possessing specialised knowledge, emotional interactions should be minimised. On the one hand, embedding industry standards for products within the dialogue system enhances information accuracy. On the other hand, communication interfaces should be optimised to enable customers to independently access product-related inspection reports, thereby reducing perceived risk. For general consumers, the fluidity of AI chatbot interactions and personalised recommendations should be prioritised. Immersive interactions can promote consumers’ flow experience, while appropriately integrating foundational knowledge dissemination balances emotional engagement with rational understanding. 5.3 Limitations and Future Research Directions This study focuses on the characteristics of sports-specialised customers as a customer group without further examining variations in their level of expertise or segmenting them on the basis of different types of sports. Customers engaged in different types of sports may have distinct professional demands for products. This singular approach to sample categorisation struggles to accurately capture customers’ differentiated needs, hence limiting the generalizability of the research conclusions. Future research could enrich the dimensions of demographic analysis by incorporating factors such as expertise level, sport type, risk preference, and brand loyalty into existing classifications. This approach would allow for a deeper exploration of how these variables influence AI interaction decisions. Furthermore, this study focuses solely on text-based AI chatbots, whereas real-world consumer ecosystems feature diverse AI applications, such as VR/AR immersive shopping experiences, virtual influencer sales, and AI recommendation systems. Different AI interaction formats have distinct influence mechanisms on consumers. Future research could conduct cross-scenario comparative analyses to examine the differential effects of various AI technologies on flow experiences, perceived risk, and impulse buying behaviour, thereby clarifying the functional boundaries of different AI technologies. Furthermore, this study relied solely on two-stage questionnaire data for analysis. While this approach validated correlations between variables, it failed to dynamically capture the evolving cognitive and emotional states of consumers during AI interactions. Future research could integrate behavioural experiments by manipulating different dimensions of AI communication quality to identify the underlying mechanisms affecting consumer psychology and behaviour. Eye-tracking experiments could also be employed to capture consumers’ emotional fluctuations and attention allocation during interactions, enabling deeper analysis of psychological mechanisms and enhancing the credibility of research conclusions. Declarations Competing interests The authors declare no competing interests. Data availability The datasets generated during or analysed during the current study are available from the corresponding author on reasonable request. Ethical approval The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of Huaqiao University (Number: HQUIRB 202512–11, Date: 26 December 2025). The scope of approval covered all research variables and data collection procedures in this study. Informed consent This study received ethical review approval on 26 December 2025, following which the electronic informed consent procedure was formally initiated. As this is a non-interventional online survey, the Institutional Review Board of Huaqiao University has granted an exemption from written informed consent. Our team obtain participants’ electronic informed consent via an online questionnaire platform. The period for collecting informed consent for the first round of questionnaires is from 27 December 2025 to 10 January 2026, following a 15-day interval, the period for collecting informed consent for the second round of questionnaires is from 26 January 2026 to 6 February 2026. The format of the electronic informed consent form is available on the questionnaire document. By clicking the informed consent confirmation button on the questionnaire’s home page, participants are deemed to have signed the informed consent form, agreeing to participate in this study and permitting the research team to collect and use the questionnaire data. All data will be anonymised and used solely for research analysis, report writing and the presentation of findings. The research team has fully informed participants of the study’s objectives, data usage protocols, anonymity safeguards and the absence of risks associated with participation. This study strictly adheres to the principle of voluntary participation; participants may refuse to participate or withdraw at any time without suffering any adverse consequences. This study does not involve vulnerable groups or minors. Author Contributions Conceptualisation, Q.M. and Q.F.; Methodology, Q.M. and Q.F; Software, H.H.; Validation, W.Y., Q.Z. and H.H.; Formal Analysis, Q.F.; Investigation, Q.F., W.Y. and Q.Z.; Resources, Q.M. and H.H.; Data Curation, Q.F. and H.H.; Writing – Original Draft Preparation, Q.F., W.Y. and Q.Z.; Writing – Review & Editing, Q.M. and H.H.; Visualisation, Q.F. and W.Y.; Supervision, Q.M.; Project Administration, Q.M.; Funding Acquisition, Q.M. and H.H. All authors have read and agreed to the published version of the manuscript. References Alalwan AA, Algharabat R, Abu El Samen A, Albanna H, Al-Okaily M (2025) Examining the impact of anthropomorphism and AI-chatbots service quality on online customer flow experience – Exploring the moderating role of telepresence. J Consumer Mark 42(4):448–471 Barrouillet P (2011) Dual-process theories and cognitive development: Advances and challenges. Dev Rev 31(2):79–85 Beatty SE, Elizabeth Ferrell M (1998) Impulse buying: Modeling its precursors. J Retail 74(2):169–191 Buechner B, Stadler Blank A, Escoe B, Blaney I (2024) Expertise matters: Why consumers are less likely to follow AI recommendations in sports betting. Int J Sports Mark Spons 26(4):731–747 Cabeza-Ramírez LJ, Sánchez-Cañizares SM, Santos-Roldán LM, Fuentes-García FJ (2022) Impact of the perceived risk in influencers' product recommendations on their followers' purchase attitudes and intention. Technol Forecast Soc Chang 184:121997 Chang YP, Zhu DH (2012) The role of perceived social capital and flow experience in building users’ continuance intention to social networking sites in China. Comput Hum Behav 28(3):995–1001 Chen C-C, Lin Y-C (2018) What drives live-stream usage intention? The perspectives of flow, entertainment, social interaction, and endorsement. Telematics Inform 35(1):293–303 Chen N, Yang Y (2023) The role of influencers in live streaming e-commerce: Influencer trust, attachment, and consumer purchase intention. J Theoretical Appl Electron Commer Res 18(3):1601–1618 Chen S, Li X, Liu K, Wang X (2023) Chatbot or human? The impact of online customer service on consumers' purchase intentions. Psychol Mark 40(11):2186–2200 Cheng Y, Jiang H (2020) How do AI-driven chatbots impact user experience? Examining gratifications, perceived privacy risk, satisfaction, loyalty, and continued use. J Broadcast Electron Media 64(4):592–614 Cheng Z, Fan W, Shao B, Jia W, Zhang Y (2024) The impact of intelligent customer service agents’ initial response on consumers’ continuous interaction intention. J Retailing Consumer Serv 76:103585 Chinchanachokchai S, Thontirawong P, Chinchanachokchai P (2021) A tale of two recommender systems: The moderating role of consumer expertise on artificial intelligence based product recommendations. J Retailing Consumer Serv 61:102528 Chiu W, Cho H, Chua H (2023) The dual roles of trust and risk in sport consumer decision-making in social commerce: An information adoption model. Sport Mark Q 32(4):267–283 Chiu W, Kim T, Won D (2018) Predicting consumers’ intention to purchase sporting goods online. Asia Pac J Mark Logistics 30(2):333–351 Chung M, Ko E, Joung H, Kim SJ (2020) Chatbot e-service and customer satisfaction regarding luxury brands. J Bus Res 117:587–595 Coelho F, Aniceto I, Bairrada CM, Silva P (2023) Personal values and impulse buying: The mediating role of hedonic shopping motivations. J Retailing Consumer Serv 72:103236 Dalvi-Esfahani M, Barati-Ahmadabadi H, Ramayah T, Turner JJ, Iahad A, N., Azar N (2025) Stimulus-organism-response framework of decision-makers intention to adopt generative AI to replace entry-level jobs: The moderating impact of personality traits. Technol Forecast Soc Chang 219:124291 Dholakia UM (2001) A motivational process model of product involvement and consumer risk perception. Eur J Mark 35(11–12):1340–1362 Feng Z, Al Mamun A, Masukujjaman M, Yang Q (2023) Modeling the significance of advertising values on online impulse buying behavior. Humanit Social Sci Commun 10(1):728 Grünzner M, Richter I, White MP, Pahl S (2025) Hold on, do I really need this? Countering impulse clothing purchases through short reflections. Clean Responsible Consum 17:100280 Hao R, Li C (2025) How AI chatbots shape satisfactory experiences: A combined perspective of competence expansion and emotional extension. Technol Forecast Soc Chang 212:123979 Jacoby J (2002) Stimulus-organism-response reconsidered: An evolutionary step in modeling (consumer) behavior. J Consumer Psychol 12(1):51–57 Kim J, Kim M, Lee S-M (2025) Unlocking trust dynamics: An exploration of playfulness, expertise, and consumer behavior in virtual influencer marketing. Int J Human–Computer Interact 41(1):378–390 Lee J-C, Xiong Ln (2024) The impact of informational antecedents on Generation Z consumers’ evaluations of cross-border e-commerce transaction costs and purchase decisions. Technol Forecast Soc Chang 205:123513 Lee M, Park J-S (2022) Do parasocial relationships and the quality of communication with AI shopping chatbots determine middle-aged women consumers' continuance usage intentions? J Consumer Behav 21(4):842–854 Longoni C, Bonezzi A, Morewedge CK (2019) Resistance to medical artificial intelligence. J Consum Res 46(4):629–650 Milli S, Lieder F, Griffiths TL (2021) A rational reinterpretation of dual-process theories. Cognition 217:104881 Nakamura J, Csikszentmihalyi M (2009) Flow theory and research. Oxford University Press, New York, NY, US Redine A, Deshpande S, Jebarajakirthy C, Surachartkumtonkun J (2022) Impulse buying: A systematic literature review and future research directions. Int J Consumer Stud 47(1):3–41 Rizomyliotis I, Kastanakis MN, Giovanis A, Konstantoulaki K, Kostopoulos I (2022) How mAy I help you today? The use of AI chatbots in small family businesses and the moderating role of customer affective commitment. J Bus Res 153:329–340 Sengoz A, Cavusoglu M, Kement U, Bayar SB (2024) Unveiling the symphony of experience: Exploring flow, inspiration, and revisit intentions among music festival attendees within the SOR model. J Retailing Consumer Serv 81:104043 Shao B, Cheng Z, Wan L, Yue J (2021) The impact of cross border E-tailer's return policy on consumer's purchase intention. J Retailing Consumer Serv 59:102367 Shin H, Bunosso I, Levine LR (2022) The influence of chatbot humour on consumer evaluations of services. Int J Consumer Stud 47(2):545–562 Song M, Xing X, Duan Y, Cohen J, Mou J (2022) Will artificial intelligence replace human customer service? The impact of communication quality and privacy risks on adoption intention. J Retailing Consumer Serv 66:102900 Spindler V, Schunk H, Könecke T (2023) Sustainable consumption in sports fashion – German runners' preference and willingness to pay for more sustainable sports apparel. Sustainable Prod Consum 42:411–422 Tan R, Li Y, Huang Q, Liu H (2025) Enhancing customer service chatbot effectiveness: The effect of dyadic communication traits on customer purchase intention. J Association Inform Syst 26(3):799–831 Trivedi J, Kasilingam D, Arora P, Soni S (2022) The effect of augmented reality in mobile applications on consumers' online impulse purchase intention: The mediating role of perceived value. J Consumer Behav 21(4):896–908 Uhm J-P, Kim S, Do C, Lee H-W (2022) How augmented reality (AR) experience affects purchase intention in sport E-commerce: Roles of perceived diagnosticity, psychological distance, and perceived risks. J Retailing Consumer Serv 67:103027 Wu I-L, Chiu M-L, Chen K-W (2020) Defining the determinants of online impulse buying through a shopping process of integrating perceived risk, expectation-confirmation model, and flow theory issues. Int J Inf Manag 52:102099 Wu Y, Xin L, Li D, Yu J, Guo J (2021) How does scarcity promotion lead to impulse purchase in the online market? A field experiment. Inf Manag 58(1):103283 Xin M, Jian L, Liu W, Bao Y (2025) Exploring the effect of live streaming atmospheric cues on consumer impulse buying: A flow experience perspective. J Theoretical Appl Electron Commer Res 20(2):149 Zhang X, Yu X (2020) The impact of perceived risk on consumers’ cross-platform buying behavior. Front Psychol 11:592246 Zhang X, Zhang S (2024) Investigating impulse purchases in live streaming e-commerce: A perspective of match-ups. Technol Forecast Soc Chang 205:123513 Zhang Y, Ding Z, Sun J, Zhao X, Hu X, Yang Z (2025) Optimizing AI strategies in e-commerce customer service: An agent-based simulation. Electron Markets 35(1):77 Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted 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-9008587","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":620659306,"identity":"e8f0e9b3-665d-4a6a-a615-d19bb0e8ff77","order_by":0,"name":"Qingguang Meng","email":"","orcid":"","institution":"Huaqiao University","correspondingAuthor":false,"prefix":"","firstName":"Qingguang","middleName":"","lastName":"Meng","suffix":""},{"id":620659307,"identity":"5e6f608b-cec5-4d10-8473-7fb227e3c4d8","order_by":1,"name":"Qiongfeng Feng","email":"","orcid":"","institution":"Huaqiao University","correspondingAuthor":false,"prefix":"","firstName":"Qiongfeng","middleName":"","lastName":"Feng","suffix":""},{"id":620659308,"identity":"e953b5e4-e0a2-478e-b931-9810089c0a43","order_by":2,"name":"Wenhui Yue","email":"","orcid":"","institution":"Huaqiao University","correspondingAuthor":false,"prefix":"","firstName":"Wenhui","middleName":"","lastName":"Yue","suffix":""},{"id":620659309,"identity":"6adad7d7-f21f-4d20-baf2-28045f03ac1d","order_by":3,"name":"Qingqing Zheng","email":"","orcid":"","institution":"Huaqiao University","correspondingAuthor":false,"prefix":"","firstName":"Qingqing","middleName":"","lastName":"Zheng","suffix":""},{"id":620659310,"identity":"42ee7b2f-99a0-453b-afc6-0ff5f4f3c117","order_by":4,"name":"Hongtian Huang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA/UlEQVRIiWNgGAWjYBACAxDB2AAjKyzAohLEaWEDkWckSNICYrQRocVcIvnZw687bPLk45sbPxfOk0jccID54G0eBrs8XFosZ6SZG8ueSSs2PMbYLD1zG0gLW7I1D0NyMU6H3Ugwk5ZsO5y4sY2xQZoXrIXHTJqH4UBiA04t6d9gWpp/884BaeH/RkBLjpnkR6CW+WyMbdK8DWBb2PBrOfOmTJqxLS1xA1timzXPMQnjmYfZjC3nGCTj1nI8fZvkzzabxPnNxx/f5qmxke073vzwxpsKO5xaQICZB6T3AITj2MAMNgqPeiBg/AEk5KGG2uNXOwpGwSgYBSMRAABpRlmDuwrrCwAAAABJRU5ErkJggg==","orcid":"","institution":"Linyi University","correspondingAuthor":true,"prefix":"","firstName":"Hongtian","middleName":"","lastName":"Huang","suffix":""}],"badges":[],"createdAt":"2026-03-02 09:54:40","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9008587/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9008587/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":106977668,"identity":"ca4ba08d-ea4b-4803-88a9-cfd3f0f895d7","added_by":"auto","created_at":"2026-04-15 11:06:34","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":47368,"visible":true,"origin":"","legend":"\u003cp\u003eResearch model\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-9008587/v1/073d47f4fc17d1a9e03888ac.png"},{"id":106994387,"identity":"e4e6edb0-7a58-4109-9516-6f146cb748e7","added_by":"auto","created_at":"2026-04-15 15:08:12","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":294048,"visible":true,"origin":"","legend":"\u003cp\u003eStandardised path coefficients in a structural equation model\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-9008587/v1/adc3baf9b7e9e5e734997bc2.png"},{"id":106994313,"identity":"5232b4fc-1231-4874-af70-4ba604a8bbad","added_by":"auto","created_at":"2026-04-15 15:07:43","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":37956,"visible":true,"origin":"","legend":"\u003cp\u003eThe moderating effect of sports-specialisedcustomers on AI chatbot communication quality and flow experience\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-9008587/v1/f1a4ef9e8918f0e8e0f61a12.png"},{"id":106977671,"identity":"f7b1eff1-ab37-4ad5-becd-41e87a5b6ca6","added_by":"auto","created_at":"2026-04-15 11:06:34","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":39528,"visible":true,"origin":"","legend":"\u003cp\u003eThe moderating effect of sports-specialised customers on AI chatbot communication quality and perceived risk\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-9008587/v1/c310baf27a9981a74b49e55e.png"},{"id":108182558,"identity":"1e7823b9-b334-490a-8108-cb61ab655422","added_by":"auto","created_at":"2026-04-30 08:59:25","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":752889,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9008587/v1/8992cddb-7393-4dbf-8fe2-56217f5b027a.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Are Professionals also influenced by AI to Make Impulse Buying? A Dual-System Cognitive Process of Sports-Specialised Customers","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eAI chatbots, leveraging 24/7 real-time responsiveness and personalised recommendation algorithms, have emerged as pivotal tools in e-commerce for enhancing sales conversion rates and optimising service efficiency(Song, Xing, Duan, Cohen, \u0026amp; Mou, 2022). The implementation of AI chatbots has not only significantly enhanced the consumer buying experience and stimulated impulse buying behaviour but also contributed to substantial revenue growth for e-commerce enterprises. However, the efficacy of AI chatbots in enhancing purchase intention among professional consumers in specialised sectors, such as sports-specialised consumers, still requires further empirical investigation. Sports training equipment is highly specialised, and consumers in the sustainable consumption lifecycle demonstrate more detailed demands for product functionalities, materials, and associated technical specifications(Spindler, Schunk, \u0026amp; K\u0026ouml;necke, 2023). Particularly among the young professional consumer demographic, a propensity for instant gratification is coupled with consumption behaviours that prioritise experiential engagement and social connectivity(Coelho, Aniceto, Bairrada, \u0026amp; Silva, 2023). In this context, e-commerce enterprises should leverage AI interaction technologies to deliver expert, real-time informational support and personalised shopping recommendations, thereby enhancing the consumer\u0026rsquo;s retail experience. Existing research has demonstrated that external stimuli such as live-streaming interactions and augmented reality can effectively trigger impulse buying(Xiaoping Zhang \u0026amp; Zhang, 2024). AI chatbots also possess the ability to influence consumer decision-making, but their impact on purchasing decisions varies depending on the service level of the AI chatbot. Crucially, even when professionals make purchases influenced by AI chatbots, they often return products later because of mismatches with expectations, resulting in high after-sales costs and operational pressures for e-commerce enterprises(Shao, Cheng, Wan, \u0026amp; Yue, 2021). Therefore, exploring the underlying mechanisms through which AI chatbots influence impulse purchasing behaviour to increase corporate sales efficiency in human\u0026ndash;machine collaborative consumption scenarios has become a hot topic of sustained theoretical interest(J.-C. Lee \u0026amp; Xiong, 2024).\u003c/p\u003e\n\u003cp\u003eThe literature has begun to focus on the positive effects of AI chatbot communication quality; however, it has overlooked the conflicting effects that may arise from different cognitive mechanisms. In theory, AI chatbots can accurately capture customers\u0026apos; intentions, preferences, and emotions during conversations(Trivedi, Kasilingam, Arora, \u0026amp; Soni, 2022), providing consumers with instant service(Song et al., 2022). Through their human-like, humorous style, AI chatbots increase consumer satisfaction with service during the shopping process, thereby increasing purchasing intention(Shin, Bunosso, \u0026amp; Levine, 2022; Tan, Li, Huang, \u0026amp; Liu, 2025). Moreover, AI chatbots\u0026rsquo; precise personalised recommendation capabilities can push products tailored to consumer needs on the basis of browsing history and preference data, sparking consumer interest and driving impulse buying(S. Chen, Li, Liu, \u0026amp; Wang, 2023). These studies implicitly assume that consumers are passive recipients of AI-generated information, thus overlooking people\u0026rsquo;s capacity for deep processing of information and autonomous decision-making. With respect to the cognitive processing of professional consumers of sporting goods, dual-systems theory posits that the human mind comprises two distinct cognitive systems: System 1 operates as an unconscious, automatic mode of thinking, relying on experience, emotions, and associative memory to generate rapid responses, whereas System 2 functions as a rational, conscious mode of thinking that requires focused attention to execute cognitive processing and self-regulation(Milli, Lieder, \u0026amp; Griffiths, 2021). During the process of AI chatbots providing communication services, emotional stimuli activate System 1 to trigger a flow experience, immersing consumers in a pleasurable state. This state reduces the cognitive responses associated with rational decision-making(Sengoz, Cavusoglu, Kement, \u0026amp; Bayar, 2024), thereby driving consumers toward the intuitive impulse buying of sports products. Cognitive stimuli activate System 2, thereby triggering perceived risk. Consumers then carefully consider potential product issues, usage risks, and after-sales guarantees(I.-L. Wu, Chiu, \u0026amp; Chen, 2020; Xiaoping Zhang \u0026amp; Zhang, 2024), which may increase their perceived risk of the product and enable rational decision-making for the impulse buying of sporting goods.\u003c/p\u003e\n\u003cp\u003eIn this study, a research framework based on S-O-R theory and dual-system theory is constructed, and AI chatbot service quality is treated as an external stimulus (S). By activating consumers\u0026rsquo; flow experiences and physiological responses to perceived risk (O), this stimulus ultimately influences consumers\u0026rsquo; impulse buying behaviour toward sporting goods (R). This study reveals a dual-path mechanism through which the communication quality of AI chatbots influences impulse buying: Flow experiences catalyse impulse buying via emotional responses, whereas perceived risk inhibits it through rational deliberation. Together, these elements constitute the decision-making logic in human‒machine collaborative consumption scenarios. The research contributions of this paper are as follows: (1) It pioneers an exploration of the mechanism through which AI (machine) influences customers\u0026rsquo; (human) online impulse buying. The study reveals that System 2 cognitive processing heightens perceived risk during consumer interactions with AI chatbots, thereby inhibiting impulse buying behaviour. These findings respond to the research call by Wu et al. (2021) regarding how rational information affects online impulse buying\u003csup\u003e\u0026nbsp;\u003c/sup\u003eby expanding the theory of human\u0026ndash;machine interaction(Y. Wu, Xin, Li, Yu, \u0026amp; Guo, 2021). (2) Comparing professional judgment with AI precision recommendations, this study pioneers the comparison of human expertise versus AI expertise in new consumption scenarios. Its examination of the dual-system cognitive process underlying impulse buying among sports professionals explains the unique decision-making of experts while responding to calls by Artem Redine et al. to explore consumer impulse buying in new domains(Redine, Deshpande, Jebarajakirthy, \u0026amp; Surachartkumtonkun, 2022).\u003c/p\u003e"},{"header":"2. Theoretical Foundations and Research Hypotheses","content":"\u003cp\u003eDual-systems theory describes distinct cognitive processes occurring in System 1 and System 2(Barrouillet, 2011). System 1 relies on intuition, consumes fewer cognitive resources, and enables rapid, instinctive responses to stimuli; System 2, in contrast, analyses based on rules, requiring greater cognitive investment and complex computations(Milli et al., 2021). When System 1 dominates, individuals tend to make intuitive decisions; conversely, when System 2 takes precedence, it drives deeper thinking and rational judgment. The stimulus\u0026ndash;organism\u0026ndash;response (S-O-R) model further elucidates how external stimuli ultimately trigger specific behaviours or responses by influencing an individual\u0026rsquo;s internal state(Jacoby, 2002). The core principle of this model emphasises that an individual\u0026rsquo;s response is not directly determined solely by external stimuli but rather results from the combined effects of environmental stimuli, individual cognition, and emotional changes(Dalvi-Esfahani et al., 2025). Flow experience refers to an individual\u0026rsquo;s holistic state of engagement during an activity, often accompanied by intense pleasure and diminished self-awareness, allowing complete immersion in the present moment(C.-C. Chen \u0026amp; Lin, 2018; Nakamura \u0026amp; Csikszentmihalyi, 2009). In a flow state, consumers\u0026rsquo; cognitive processing shifts to the System 1-dominant mode outlined in dual-system theory. Consumers tend to process stimuli delivered by AI chatbots intuitively and automatically, ultimately responding to such stimuli with impulsive buying behaviour. In contrast, under perceived risk conditions, consumers\u0026rsquo; cognitive patterns shift toward the System 2-dominant mode. When confronted with information or recommendations from AI chatbots, consumers tend to evaluate potential risks through rational thinking(Cabeza-Ram\u0026iacute;rez, S\u0026aacute;nchez-Ca\u0026ntilde;izares, Santos-Rold\u0026aacute;n, \u0026amp; Fuentes-Garc\u0026iacute;a, 2022), ultimately employing risk-averse decision-making logic to reduce impulsive buying behaviour.\u003c/p\u003e\n\u003cp\u003eAdditionally, when consumers are confronted with external stimuli from AI chatbots, their identity characteristics may also influence their cognitive processing pathways. Consumers possessing specialised knowledge can leverage their understanding of products and domains to more accurately identify potential consequences that AI chatbot stimuli may trigger. Therefore, this study integrates dual-system theory with the S-O-R model to examine the influence mechanisms of flow experiences and perceived risk on impulse buying behaviour toward sports products while further exploring the moderating role of sports-specialised consumers\u0026rsquo; identity.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e2.1 Communication Quality of AI Chatbots and Impulse Buying: The Mediating Role of Flow Experience\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe S-O-R model indicates that external environmental stimuli act upon an individual\u0026rsquo;s psychological and cognitive processes, ultimately triggering specific behavioural responses(Dalvi-Esfahani et al., 2025; Feng, Al Mamun, Masukujjaman, \u0026amp; Yang, 2023). In e-commerce shopping scenarios, AI chatbots can be regarded as external stimulus variables. Leveraging communication advantages such as instant responses and human-like interactions\u003csup\u003e\u0026nbsp;\u003c/sup\u003ecan reduce consumers\u0026rsquo; cognitive processing(Song et al., 2022). On the one hand, AI chatbots offer round-the-clock automated communication that can respond to consumer inquiries at any time, thereby effectively reducing waiting periods. This process not only diminishes users\u0026rsquo; perception of time passing but also decreases the cognitive processing involved in actively searching for and analysing information, making it easier for consumers to enter a flow experience state dominated by intuition. Research has indicated that the immediacy and convenience of interacting with AI bots are key technological features that drive consumer immersion in online retail scenarios(Z. Cheng, Fan, Shao, Jia, \u0026amp; Zhang, 2024). In this low-cognitive-load communication environment, consumers are more inclined to rely on intuitive cognition to process information, thereby increasing the likelihood of experiencing flow states.\u003c/p\u003e\n\u003cp\u003eOn the other hand, with the advancement of machine learning technology, AI chatbots can construct accurate user profiles based on customer retention data, precisely identify user needs, and provide appropriate responses(Rizomyliotis, Kastanakis, Giovanis, Konstantoulaki, \u0026amp; Kostopoulos, 2022), thereby providing a more tailored interactive experience for consumers. This anthropomorphic interaction not only enhances the approachability of the service but also allows users to experience emotional support and pleasure during communication(Alalwan, Algharabat, Abu El Samen, Albanna, \u0026amp; Al-Okaily, 2025), thereby strengthening their flow experience throughout the interaction process. Moreover, the personalised recommendations provided by AI chatbots can further enhance consumer engagement, keeping them highly focused throughout the information-gathering and communication process(S. Chen et al., 2023). These high-quality communication features of AI chatbots reduce cognitive resistance among consumers, making them more likely to ignore external distractions and fully immerse themselves in interactions with the AI chatbot, thereby triggering a flow experience. On this basis, the following hypothesis is proposed in this study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eH1\u003c/strong\u003e\u003cstrong\u003e:\u003c/strong\u003e\u003cem\u003e\u0026nbsp;\u003c/em\u003eThe communication quality of AI chatbots has a positive effect on flow experiences.\u003c/p\u003e\n\u003cp\u003eDual-system theory posits that individuals possess two distinct cognitive processing systems. System 1 represents an intuitive cognitive processing mode characterised by low cognitive load and reliance on emotion-driven mechanisms(Barrouillet, 2011; Milli et al., 2021). The core characteristic of flow experience lies in the individual\u0026rsquo;s complete immersion in the present activity and diminished self-awareness(Nakamura \u0026amp; Csikszentmihalyi, 2009), a psychological state that aligns with System 1\u0026rsquo;s cognitive processing mode. Specifically, consumers in a flow state experience heightened pleasure, and immersion in this emotional state weakens their capacity for behavioural reflection, thus making them more inclined to immediate, emotional responses to external stimuli. Within this cognitive context, System 1 processing dominates, automating consumers\u0026rsquo; cognitive processing of information\u0026mdash;such as instant communication services and promotional offers\u0026mdash;provided by AI chatbots. This state reduces consumers\u0026rsquo; rational evaluation of information, making them more susceptible to unplanned impulse buying. Furthermore, the pleasure and enjoyment accompanying flow experiences trigger an affective cognitive processing mode(Sengoz et al., 2024; Xiaoping Zhang \u0026amp; Zhang, 2024). In this state, consumers unconsciously project these emotions onto their current consumption context and the products they encounter. This emotion-driven cognitive processing significantly reduces consumers\u0026rsquo; perception of potential product risks(Xiaoping Zhang \u0026amp; Zhang, 2024), weakens their rational risk-averse tendencies, and further catalyses impulsive buying decisions.\u003c/p\u003e\n\u003cp\u003eNotably, during interactions with AI chatbots, multiple external stimuli\u0026mdash;such as automatic coupon pushes, limited-time promotion pop-ups, and human-like communication tones\u0026mdash;mutually reinforce consumers\u0026rsquo; flow experiences. These stimuli reduce cognitive processing, deepen the flow experience, and prompt consumers to base their judgments of product value on immediate emotional responses. Ultimately, this influence leads to impulsive buying behaviour. On this basis, the following hypothesis is proposed in this study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eH2\u003c/strong\u003e\u003cstrong\u003e:\u003c/strong\u003e Flow experiences have a positive effect on impulse buying.\u003c/p\u003e\n\u003cp\u003eIn combination, H1 and H2 suggest that the communication quality of AI chatbots enhances consumers\u0026rsquo; flow experience, which in turn further drives their impulse buying behaviour. On this basis, the following hypothesis is proposed in this study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eH3:\u0026nbsp;\u003c/strong\u003eFlow experience mediates the relationship between communication quality in AI chatbots and impulse buying.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003e2.2 Communication Quality of AI Chatbots and Impulse Buying: The Mediating Role of Perceived Risk\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAccording to dual-system theory, when consumers encounter complex or potentially threatening external stimuli, their cognitive processing mode shifts to System 2, which requires rational analysis(Barrouillet, 2011; Milli et al., 2021). During professional consultations about sporting goods, the limitations of AI chatbot responses and their collection of private information drive consumers into the rational cognitive processing mode of System 2, thereby amplifying their perceived risks associated with sporting goods. Specifically, consumers place high demands on safety and functionality when they purchase sporting goods(Chiu, Kim, \u0026amp; Won, 2018). Although AI chatbots can provide instant communication services(Z. Cheng et al., 2024), their performance in e-commerce contexts often fails to meet consumers\u0026rsquo; specialised needs and may even exacerbate negative perceptions(Y. Zhang et al., 2025). A seamless conversational experience tends to increase consumers\u0026rsquo; expectations regarding the communication capabilities of AI chatbots. When these AI chatbots fail to address complex inquiries specific to particular sports scenarios, this cognitive dissonance can erode consumer trust in the AI chatbot. Consequently, consumers may rationally evaluate the accuracy and completeness of the information provided by the AI chatbot, which ultimately heightens their perception of the risk associated with the product.\u003c/p\u003e\n\u003cp\u003eAdditionally, while AI chatbots can provide personalized recommendations\u003csup\u003e10\u003c/sup\u003e, they still collect users\u0026rsquo; personal information\u0026mdash;such as exercise preferences, purchase history, and physical data\u0026mdash;during communication to achieve more precise conversations and recommend suitable sports products. Research has confirmed that the collection of private information triggers users\u0026rsquo; risk perception(Y. Cheng \u0026amp; Jiang, 2020). Once consumers detect in-depth collection and analysis of such personal data during interactions, this recognition activates the rational cognitive processing of System 2, prompting them to weigh the convenience of personalised services against the potential risks of data breaches. This cognitive process amplifies consumers\u0026rsquo; trust concerns about AI chatbot services, thereby increasing perceived risk(N. Chen \u0026amp; Yang, 2023; Chiu, Cho, \u0026amp; Chua, 2023). On this basis, the following hypothesis is proposed in this study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eH4:\u003c/strong\u003e\u003cem\u003e\u0026nbsp;\u003c/em\u003eThe communication quality of AI chatbots has a positive effect on perceived risk.\u003c/p\u003e\n\u003cp\u003ePerceived risk refers to an individual\u0026rsquo;s subjective assessment of the uncertainties associated with a specific product and the potential negative consequences they may trigger(Dholakia, 2001; Xiaoxue Zhang \u0026amp; Yu, 2020). System 2 cognitive processing further translates this assessment into concrete risk-avoidance strategies. Impulse buying is essentially a hedonistic, immediate behavioural response directly triggered by external stimuli(I.-L. Wu et al., 2020), while high levels of perceived risk inhibit this behaviour through various mechanisms. On the one hand, perceived risk prompts consumers to reflect on the suitability of the product for their needs and their own purchasing desires(Gr\u0026uuml;nzner, Richter, White, \u0026amp; Pahl, 2025), which compels them to continuously allocate limited cognitive resources toward risk assessment and thereby dampens immediate buying impulses(Cabeza-Ram\u0026iacute;rez et al., 2022). On the other hand, high levels of perceived risk drive consumers to rationally weigh the actual utility of the product against potential return costs, thus effectively curbing impulsive buying behaviour.\u003c/p\u003e\n\u003cp\u003eThe use of sporting goods is directly linked to consumers\u0026rsquo; physical health and safety. The safety and functional characteristics of sporting goods mean that consumers are more risk-sensitive toward these products than toward ordinary products. Therefore, even when enticed by marketing stimuli such as time-limited discounts and pop-up ads pushed by AI chatbots in e-commerce contexts(Z. Cheng et al., 2024), consumers\u0026rsquo; concerns about information asymmetry regarding products, worries about after-sales guarantees(Chiu et al., 2023; Y. Zhang et al., 2025), and doubts about the reliability of AI chatbot services will still be processed rationally by System 2 into risk perceptions. This perceived risk ultimately negatively influences impulsive buying behaviour. On this basis, the following hypothesis is proposed in this study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eH5:\u0026nbsp;\u003c/strong\u003ePerceived risk has a negative effect on impulse buying.\u003c/p\u003e\n\u003cp\u003eIn combination, H4 and H5 suggest that the communication quality of AI chatbots increases consumers\u0026rsquo; perceived risk, which in turn further weakens their impulse buying behaviour. Therefore, this study proposes the following hypothesis.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eH6:\u0026nbsp;\u003c/strong\u003ePerceived risk mediates the relationship between AI chatbot communication quality and impulse buying.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003e2.3 The Moderating Role of Sports-Specialised Consumers\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFlow experience is a psychological state in which an individual\u0026rsquo;s attention is absorbed and immersed in an activity(Nakamura \u0026amp; Csikszentmihalyi, 2009), dominated by intuitive System 1 cognitive processing. When customers with a sports-specialised background interact with AI chatbots, they do not passively receive information but instead process the messages conveyed by the AI chatbot. This process keeps System 2 continuously engaged, prompting rational analysis of the external stimuli generated by the AI chatbot(Milli et al., 2021). Research has confirmed that professional consumers possess more complex knowledge structures in their memory, enabling them to analyse and process product information more deeply than novice consumers do(Chinchanachokchai, Thontirawong, \u0026amp; Chinchanachokchai, 2021). This understanding enables sports-specialised customers to evaluate product information or recommendations provided by AI chatbots on the basis of their own expertise. They are not swayed by the superficial fluency of AI conversations or marketing rhetoric, nor do they blindly accept the chatbot\u0026rsquo;s suggestions(Buechner, Stadler Blank, Escoe, \u0026amp; Blaney, 2024; Kim, Kim, \u0026amp; Lee, 2025). This rational approach diminishes the ability of the AI chatbot to enhance flow experiences through communication quality.\u003c/p\u003e\n\u003cp\u003eFurthermore, flow experiences encompass emotional elements, such as pleasure and enjoyment(Sengoz et al., 2024). AI chatbots often employ anthropomorphic expressions and emotional interactions to enhance engagement(Song et al., 2022), thereby stimulating positive user emotions and immersion. In sports product purchasing decisions, sports-specialised customers place greater emphasis on the reliability of their own assessments of product quality and performance(Uhm, Kim, Do, \u0026amp; Lee, 2022). Consequently, they expect AI chatbots to deliver precise, professional, and practical information to help them make informed product decisions. If an AI chatbot\u0026rsquo;s communication approach emphasises emotional interaction over substantive professional support, it will not only fail to attract professionals but may even be perceived as disruptive or unprofessional. Consequently, in sports product e-commerce purchasing scenarios, the formation of flow experiences during interactions between sports-specialised customers and AI chatbots will be inhibited. On this basis, the following hypothesis is proposed in this study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eH7:\u003c/strong\u003e\u003cem\u003e\u0026nbsp;\u003c/em\u003eSports-specialised customers moderate the relationship between AI chatbot communication quality and flow experience, with sports-specialised customers weakening the positive influence of AI chatbot communication quality on the flow experience.\u003c/p\u003e\n\u003cp\u003eAccording to dual-system theory, sports-specialised customers\u0026rsquo; knowledge base leads them to rely more heavily on the rational analytical pathway of System 2 when evaluating the communication quality of AI chatbots(Barrouillet, 2011; Milli et al., 2021). This cognitive processing model enables it to keenly identify implicit information biases in AI conversations, thereby enhancing the recognition of perceived risks. Even if AI chatbots appear to function flawlessly during communication, they can still trigger doubts among sports-specialised customers about the authenticity of information and the suitability of product recommendations. Such perceptions lead customers to reject AI-generated suggestions, thereby amplifying perceived risk. Compared with general consumers, sports-specialised customers have higher expectations for the quality and performance of athletic products(Spindler et al., 2023). Their evaluation of AI communication quality centres on the accuracy of the content conveyed rather than solely focusing on the interactive experience. Therefore, when AI chatbots attempt to enhance service quality through personalised recommendations, sports-specialised customers will verify the validity of these suggestions on the basis of their own expertise(Chinchanachokchai et al., 2021). If they find that the recommended products deviate from their actual athletic needs, they will not only reject the information provided by the AI chatbot but also perceive this proactive service as misleading. This evaluation further amplifies their perception of risk with respect to product functionality.\u003c/p\u003e\n\u003cp\u003eAdditionally, in online shopping scenarios, consumers cannot physically touch or experience products. AI chatbots primarily convey information through text, making it difficult to enhance the virtual experience through immersive design and multisensory displays such as AR technology(Uhm et al., 2022). This limitation makes it challenging for sports-specialised customers\u0026mdash;who pay close attention to product details\u0026mdash;to verify actual functionality on the basis solely of textual descriptions. As a result, they are highly prone to distrust information provided by AI chatbots, which ultimately exacerbates the perception of risk. On this basis, the following hypothesis is proposed in this study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eH8:\u0026nbsp;\u003c/strong\u003eSports-specialised customers moderate the relationship between AI chatbot communication quality and perceived risk. Sports-specialised customers enhance the positive influence of AI chatbot communication quality on perceived risk.\u003c/p\u003e\n\u003cp\u003eBased on the above assumptions, the conceptual model of this study is shown in Fig 1.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFig 1.\u0026nbsp;\u003c/strong\u003eResearch model\u003c/p\u003e"},{"header":"3. Data and Methods","content":"\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e\n \u003ch2\u003e3.1 Sample and Data Collection\u003c/h2\u003e\n \u003cp\u003eTo ensure that participants had a thorough understanding of the survey content and to guarantee the reliability and validity of the data, this study targeted consumers with online shopping experience. Specifically, a portion of the sample was restricted to individuals with professional knowledge and skills in the sports field. To minimise common method bias during questionnaire distribution, this study employed a two-stage time-lagged design for data collection. The first phase involved the distribution of questionnaires via the Tencent survey platform, primarily measuring respondents\u0026rsquo; demographic characteristics, such as gender, age, years of online shopping experience, and online shopping frequency. Additionally, it assessed the independent variable of communication quality with AI chatbots and the moderating variable of sports- specialised customers. A total of 527 questionnaires were distributed during this phase. After invalid samples were excluded, 432 valid questionnaires were ultimately collected. Two weeks after the first round of surveys concluded, we invited valid participants who had completed Phase One to participate in Phase Two. In this phase, the questionnaire included core variable measurement items such as flow experience, perceived risk, and impulse buying behaviour.\u003c/p\u003e\n \u003cp\u003eIn the second phase, a total of 432 questionnaires were distributed. After further screening for invalid responses, 304 valid questionnaires were ultimately obtained. To ensure the precise matching of data across both phases while strictly protecting participant privacy, we requested that participants provide the last four digits of their mobile phone number as an anonymous identifier in both phases. Full contact details were immediately deleted after data cleaning and matching to guarantee that no traceable personal identification remained. Furthermore, all the questionnaires prominently displayed the research-informed consent form on the first page, detailing the study objectives, data usage, and confidentiality measures. Participants only proceeded to the formal questionnaire after providing their consent.\u003c/p\u003e\n \u003cp\u003eAmong the valid samples, males accounted for 56.91% and females for 43.09%, indicating a male-dominated sample with a relatively balanced gender ratio. The participants were predominantly young and middle-aged adults, with the greatest proportion (32.57%) aged 31\u0026ndash;35, followed by those aged 26\u0026ndash;30 (26.97%). Those aged 21\u0026ndash;25 accounted for 19.08%, while those aged 36 and above accounted for 21.38%. This age structure indicates that the core sample consisted of mature users with strong purchasing power. The vast majority of participants had more than one year of online shopping experience: 38.16% had 1\u0026ndash;3 years, 28.29% had 3\u0026ndash;5 years, and 22.70% had more than 5 years. This indicates the sample\u0026rsquo;s overall familiarity with online shopping processes and environments, which enhanced their comprehension of survey content and the quality of their responses. Nearly three-quarters of the respondents shopped online 3 or more times per month, with 3\u0026ndash;5 times being the most common at 37.17%, followed by 6\u0026ndash;10 times at 26.32%, and more than 10 times at 10.20%. This indicates that sample users generally exhibited active online consumption habits, meeting the study\u0026rsquo;s screening requirement for \u0026ldquo;having online shopping experience.\u0026rdquo;\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\n \u003ch2\u003e3.2 Measurement Items\u003c/h2\u003e\n \u003cp\u003eAll the variables in this research model were selected from established scales published in authoritative academic journals. During the translation process, we implemented a rigorous back-translation procedure conducted collaboratively by bilingual experts. Through this process, items exhibiting semantic discrepancies were discussed and revised, ultimately resulting in the formal survey questionnaire used in this study.\u003c/p\u003e\n \u003cdiv id=\"Sec9\" class=\"Section3\"\u003e\n \u003ch2\u003e3.2.1 Dependent Variable\u003c/h2\u003e\n \u003cp\u003eThe measurement of impulse buying was primarily based on the research by Beatty and Elizabeth (1998) and incorporated measurement items from Xin et al. (\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2025\u003c/span\u003e)(Beatty \u0026amp; Elizabeth Ferrell, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e1998\u003c/span\u003e; Xin, Jian, Liu, \u0026amp; Bao, \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). The instrument comprises three items, with a typical example being \u0026ldquo;I am someone who buys products that weren\u0026rsquo;t originally planned.\u0026rdquo;\u003c/p\u003e\n \u003c/div\u003e\n \u003cdiv id=\"Sec10\" class=\"Section3\"\u003e\n \u003ch2\u003e3.2.2 Independent Variable\u003c/h2\u003e\n \u003cp\u003eThe measurement of AI chatbot communication quality employed a scale developed by Chung et al. (\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2020\u003c/span\u003e), drawing upon the research of Lee and Park (\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2022\u003c/span\u003e)(Chung, Ko, Joung, \u0026amp; Kim, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; M. Lee \u0026amp; Park, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). It comprises 12 items, with a typical example being \u0026ldquo;When shopping online, my communication with the AI chatbot is timely.\u0026rdquo;\u003c/p\u003e\n \u003c/div\u003e\n \u003cdiv id=\"Sec11\" class=\"Section3\"\u003e\n \u003ch2\u003e3.2.3 Mediating Variables\u003c/h2\u003e\n \u003cp\u003eThe measurement of flow experience employed the scale developed by Chang and Zhu (\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2012\u003c/span\u003e)(Chang \u0026amp; Zhu, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2012\u003c/span\u003e), drawing upon research by Chen and Lin (\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2018\u003c/span\u003e) et al.(C.-C. Chen \u0026amp; Lin, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). It comprises four items, with a typical example being \u0026ldquo;When shopping online, I feel time passes quickly when communicating with the AI chatbot.\u0026rdquo; Perceived risk was measured on the basis of Dholakia\u0026rsquo;s (\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2001\u003c/span\u003e) study(Dholakia, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2001\u003c/span\u003e), incorporating items from Cabeza-Ram\u0026iacute;rez et al. (\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2022\u003c/span\u003e)(Cabeza-Ram\u0026iacute;rez et al., \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). It comprises three items, with a representative example being: \u0026ldquo;It\u0026rsquo;s risky to buy products recommended/promoted by AI chatbots.\u0026rdquo;\u003c/p\u003e\n \u003c/div\u003e\n \u003cdiv id=\"Sec12\" class=\"Section3\"\u003e\n \u003ch2\u003e3.2.4 Regulating Variables\u003c/h2\u003e\n \u003cp\u003eThe moderating variable in this study was sports-specialised customers. To avoid suggestive bias from direct questioning in the questionnaire design, we measured this variable with a background screening item: \u0026ldquo;What is your professional field?\u0026rdquo; The options were as follows: \u0026ldquo;Sports field: holding nationally certified sports qualifications, possessing a higher education background in sports, or currently engaged in sports-related professional work,\u0026rdquo; \u0026ldquo;Science, Engineering, Agriculture, and Medicine: e.g., computer science, engineering, medicine, biology,\u0026rdquo; \u0026ldquo;Humanities and Social Sciences: e.g., literature, history, economics, law, arts,\u0026rdquo; \u0026ldquo;Other.\u0026rdquo; In subsequent data analysis, respondents selecting the sports field were coded as 1, indicating a professional sports background. All other respondents were coded as 0 and served as the control group. This approach concealed the purpose of the study while effectively distinguishing key moderating variables.\u003c/p\u003e\n \u003c/div\u003e\n \u003cdiv id=\"Sec13\" class=\"Section3\"\u003e\n \u003ch2\u003e3.2.5 Control Variables\u003c/h2\u003e\n \u003cp\u003eThis study controlled for gender, age, years of online shopping experience, and online shopping frequency. Except for the control variables and moderating variable, other variables were measured on a five-point Likert scale, where 1 denotes strongly disagree and 5 denotes strongly agree.\u003c/p\u003e\n \u003c/div\u003e\n\u003c/div\u003e"},{"header":"4. Results","content":"\u003cp\u003e\u003cstrong\u003e4.1 Common Method Variance and Collinearity Test\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study employed the Harman single-factor test to examine common method bias. The results revealed that the total variance explained by the sample data was 65.166%. The variance contribution rate of the first principal component obtained without rotation was 36.427%, which was below the 40% critical threshold. This result suggests that the sample did not exhibit severe common method bias. Collinearity testing was performed on the sample data using SPSS. As shown in Table 3, the highest VIF (1.384) among the variables was less than 5, indicating that no severe multicollinearity issues existed in the sample data.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003e4.2 Reliability and Validity Analysis\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIn this study, Cronbach\u0026rsquo;s \u0026alpha; coefficient was used to assess the reliability of the measurement scales for each variable. The results revealed that the Cronbach\u0026rsquo;s \u0026alpha; values for AI chatbot communication quality, flow experience, perceived risk, and impulse buying behaviour were 0.940, 0.837, 0.814, and 0.773, respectively, with all the measurement items meeting the standard. Amos software was used to perform confirmatory factor analysis, with \u0026chi;\u0026sup2;/df, SRMR, RMSEA, NFI, TLI, and CFI as the core fit indices. The detailed results are presented in Table 1. Model comparisons revealed that as the number of factors decreased, all fit indices deteriorated. The four-factor model demonstrated optimal fit (\u0026chi;\u0026sup2;/df = 1.134, SRMR = 0.036, RMSEA = 0.021, NFI = 0.939, TLI = 0.991, CFI = 0.992). As indicated by Table 2, the square root of the average absolute value (AVE) consistently exceeds the absolute value of the Pearson correlation coefficients between variables, confirming the discriminant validity among these variables.\u003c/p\u003e\n\u003cp\u003eTable 1. Results of confirmatory factor analysis (N=304)\u003c/p\u003e\n\u003cdiv\u003e\n \u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"82%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 22px;\"\u003e\n \u003cp\u003eModel\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e\u0026chi;\u0026sup2;/df\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003eSRMR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003eRMSEA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003eNFI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003eTLI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003eCFI\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 22px;\"\u003e\n \u003cp\u003eFour-factor model\u003c/p\u003e\n \u003cp\u003e（AIC, FE, PR, IB）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e1.134\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0.036\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e0.021\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e0.939\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e0.991\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e0.992\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 22px;\"\u003e\n \u003cp\u003eThree-factor model\u003c/p\u003e\n \u003cp\u003e（AIC, FE+PR, IB）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e2.847\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0.094\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e0.078\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e0.845\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e0.880\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e0.893\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 22px;\"\u003e\n \u003cp\u003eTwo-factor model\u003c/p\u003e\n \u003cp\u003e（AIC, FE+PR+IB）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e3.809\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0.110\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e0.096\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e0.790\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e0.817\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e0.835\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 22px;\"\u003e\n \u003cp\u003eOne-factor model\u003c/p\u003e\n \u003cp\u003e（AIC+FE+PR+IB）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e5.847\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13px;\"\u003e\n \u003cp\u003e0.128\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e0.126\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e0.676\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e0.684\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e0.714\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003eNotes: AIC=AI chatbot communication quality; FE=flow experience; PR=perceived risk; IB=impulse buying\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003e4.3 Descriptive Statistics and Correlation Analysis\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo ensure that the applicability conditions for subsequent parameter tests are met, SPSS statistical software was used in this study to calculate the skewness and kurtosis coefficients of each measurement item, thereby examining the normal distribution characteristics of the sample data. The results revealed that the absolute values of skewness for all the measurement items in this study did not exceed the critical threshold of 3 and that the absolute values of the kurtosis coefficients were all less than 8. On this basis, the sample data can be deemed to meet the requirements for an approximate normal distribution. Furthermore, the VIF values for all the variables were less than 2, indicating no severe multicollinearity issues. The descriptive statistics and correlation analysis results are presented in Table 2. As shown in Table 2, all correlation coefficients fall within reasonable ranges. Specifically, AI chatbot communication quality was significantly positively correlated with flow experience (r = 0.270, p \u0026lt; 0.001) and with perceived risk (r = 0.275, p \u0026lt; 0.001). Flow experience was significantly positively correlated with impulse buying behaviour (r = 0.327, p \u0026lt; 0.001), whereas perceived risk was significantly negatively correlated with impulse buying behaviour (r = -0.219, p \u0026lt; 0.001). These findings provide preliminary validation of the research hypotheses.\u003c/p\u003e\n\u003cp\u003eTable 2. Descriptive statistics and Pearson correlation analysis (N=304)\u003c/p\u003e\n\u003cdiv\u003e\n \u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"627\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e2\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e3\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e4\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e5\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e6\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e7\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e8\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e9\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e1. Gender\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e2. Age\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e-0.178**\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e3. Years of Online Shopping\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.014\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.316***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47px;\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e4. Online Shopping Frequency\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e-0.019\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.077\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47px;\"\u003e\n \u003cp\u003e-0.040\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47px;\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e5. Sports-Specialised Customers\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e-0.053\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e-0.012\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47px;\"\u003e\n \u003cp\u003e-0.013\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47px;\"\u003e\n \u003cp\u003e-0.057\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e6.AI Chatbot Communication Quality\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e-0.109\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.003\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47px;\"\u003e\n \u003cp\u003e-0.057\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47px;\"\u003e\n \u003cp\u003e-0.030\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.292***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e(0.758)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e7. Flow Experience\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.109\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e-0.029\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47px;\"\u003e\n \u003cp\u003e-0.059\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47px;\"\u003e\n \u003cp\u003e-0.046\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.326***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.270***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e(0.752)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e8. Perceived Risk\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e-0.055\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e-0.002\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47px;\"\u003e\n \u003cp\u003e0.051\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47px;\"\u003e\n \u003cp\u003e-0.021\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e-0.127*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.275***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.195***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e(0.772)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e9. Impulse Buying\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.070\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.021\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47px;\"\u003e\n \u003cp\u003e-0.021\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47px;\"\u003e\n \u003cp\u003e-0.072\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.245***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.252***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.327***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e-0.219***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e(0.731)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003eMean\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e1.398\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e2.645\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47px;\"\u003e\n \u003cp\u003e2.628\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47px;\"\u003e\n \u003cp\u003e2.204\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.470\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e2.662\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e2.696\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e2.618\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e3.582\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003eSD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.490\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e1.171\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47px;\"\u003e\n \u003cp\u003e0.953\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47px;\"\u003e\n \u003cp\u003e0.946\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.500\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e1.039\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e1.160\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e1.200\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e1.019\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003eVIF\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e1.087\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e1.169\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47px;\"\u003e\n \u003cp\u003e1.140\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47px;\"\u003e\n \u003cp\u003e1.020\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e1.288\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e1.355\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e1.384\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e1.368\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e1.346\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003eNotes: * p\u0026lt;0.05, ** p\u0026lt;0.01, *** p\u0026lt;0.001; The numbers in diagonal brackets represent the square root of the AVE value.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e4.4 Hypothesis Testing\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e4.4.1 Direct Effect Testing\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIn this study, SPSS was employed to examine the mechanism through which AI chatbot communication quality influences impulsive buying behaviour via two distinct pathways. As shown by the regression analysis results in Table 4, in Model 1, AI chatbot communication quality significantly and positively affected flow experience (\u0026beta; = 0.281, p \u0026lt; 0.001), validating H1. In Model 8, flow experience significantly and positively influenced impulsive buying behaviour (\u0026beta; = 0.267, p \u0026lt; 0.001), validating H2. In Model 4, AI chatbot communication quality significantly and positively influenced perceived risk (\u0026beta; = 0.276, p \u0026lt; 0.001), validating H4. In Model 9, perceived risk exerted a significant negative effect on impulsive buying behaviour (\u0026beta; = -0.310, p \u0026lt; 0.001), validating H5.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e4.4.2 Mediation Effect Test\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study employed hierarchical regression analysis to separately examine the mediating effects of flow experience and perceived risk. As shown in Table 4, after flow experience was incorporated as a mediating variable in Model 8, AI chatbot communication quality still exerted a significant positive influence on impulse buying behaviour (\u0026beta; = 0.185, p \u0026lt; 0.001). However, compared with Model 7, the regression coefficient for AI chatbot communication quality decreased from 0.260 to 0.185. These findings indicate that flow experience mediated the relationship between AI chatbot communication quality and impulsive buying behaviour, validating Hypothesis H3. After perceived risk was incorporated as a mediating variable in Model 9, AI chatbot communication quality still exerted a significant positive effect on impulsive buying behaviour (\u0026beta; = 0.346, p \u0026lt; 0.001), with its regression coefficient increasing from 0.260 in Model 7 to 0.346. Moreover, perceived risk had a significant negative effect on impulsive buying behaviour (\u0026beta; = -0.310, p \u0026lt; 0.001). These findings indicate that perceived risk moderated the true relationship between AI chatbot communication quality and impulsive buying behaviour, validating Hypothesis H6.\u003c/p\u003e\n\u003cp\u003eThe bootstrap test results for the mediating effect (based on 5,000 samples) further revealed the underlying mechanism through which AI chatbot communication quality influences impulse buying. As shown in Table 3, the total effect of AI chatbot communication quality on impulse buying was 0.255, whereas its indirect effect via flow experience was 0.074. The 95% confidence interval did not contain zero, indicating that flow experience mediated the relationship between AI chatbot communication quality and impulse buying, thus validating H3. The indirect effect of AI chatbot communication quality on impulse buying via perceived risk was -0.084, with the 95% confidence interval not containing zero. These findings indicate that perceived risk mediated the relationship between AI chatbot communication quality and impulse buying, further validating H6.\u003c/p\u003e\n\u003cp\u003eTable 3. Mediating effect test of the bootstrap method (N=304)\u003c/p\u003e\n\u003cdiv\u003e\n \u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"538\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003ctd nowrap=\"\" colspan=\"2\" rowspan=\"2\" style=\"width: 66px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePath\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" rowspan=\"2\" style=\"width: 62px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eEffect\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" rowspan=\"2\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSE\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" colspan=\"2\" style=\"width: 60px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e95%CI\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eLLCI\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 60px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eULCI\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd nowrap=\"\" colspan=\"2\" style=\"width: 66px;\"\u003e\n \u003cp\u003eTotal effect\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 62px;\"\u003e\n \u003cp\u003e0.255\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.055\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.147\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 60px;\"\u003e\n \u003cp\u003e0.363\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd nowrap=\"\" rowspan=\"2\" style=\"width: 265px;\"\u003e\n \u003cp\u003eAI chatbot communication quality\u0026rarr; Flow experience\u0026rarr;\u0026nbsp;Sports-specialized consumers\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003eIndirect effect\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 62px;\"\u003e\n \u003cp\u003e0.074\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.036\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 60px;\"\u003e\n \u003cp\u003e0.124\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003eDirect effect\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 62px;\"\u003e\n \u003cp\u003e0.182\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.055\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.073\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 60px;\"\u003e\n \u003cp\u003e0.290\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd nowrap=\"\" rowspan=\"2\" style=\"width: 265px;\"\u003e\n \u003cp\u003eAI chatbot communication quality\u0026rarr; Perceived risk\u0026rarr;\u0026nbsp;Sports-specialized consumers\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003eIndirect effect\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 62px;\"\u003e\n \u003cp\u003e-0.084\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e-0.135\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 60px;\"\u003e\n \u003cp\u003e-0.044\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003eDirect effect\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 62px;\"\u003e\n \u003cp\u003e0.339\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 53px;\"\u003e\n \u003cp\u003e0.054\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.232\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 60px;\"\u003e\n \u003cp\u003e0.446\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003eIn this study, Amos software was used to analyse the structural equation model. The overall model fit indices were as follows: \u0026chi;\u0026sup2;/df = 1.252, RMSEA = 0.029, GFI = 0.931, TLI = 0.984, CFI = 0.985, with all metrics meeting the ideal standards. The standardised path coefficients of the structural equation model and the significance levels of the relationships between the variables are presented in Fig 2. The results fully align with the hypotheses proposed in this study, further validating the relationships among AI chatbot communication quality, flow experience, perceived risk, and impulse buying.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFig 2.\u0026nbsp;\u003c/strong\u003eStandardised path coefficients in a structural equation model\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e4.4.3 Moderation Effect Test\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe moderator variable \u0026ldquo;sports-specialised customers\u0026rdquo; in this study involved categorical data. Therefore, before conducting stratified regression analysis, we first performed dummy variable coding. Simultaneously, the independent variable \u0026ldquo;AI chatbot communication quality\u0026rdquo; underwent centring. On the basis of the processed independent and moderator variables, we subsequently constructed their interaction term. As shown in Table 4, in Model 3, AI chatbot communication quality significantly positively influenced flow experience (\u0026beta; = 0.317, p \u0026lt; 0.001), whereas the moderating effect of sports-specialised customers was significantly negative (\u0026beta; = -0.163, p \u0026lt; 0.05). This finding indicates that the presence of sports-specialised customers weakened the positive effect of AI chatbot communication quality on flow experience, validating H7. In Model 6, AI chatbot communication quality significantly and positively influenced perceived risk (\u0026beta; = 0.199, p \u0026lt; 0.05). Concurrently, the moderating effect of sports-specialised customers was significantly positive (\u0026beta; = 0.201, p \u0026lt; 0.01), indicating that the presence of sports-specialised customers amplified the positive effect of AI chatbot communication quality on perceived risk. Thus, H8 is validated. The moderating effect of sports-specialised customers is illustrated in Fig 3 and 4.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFig\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003e3\u003c/strong\u003e\u003cstrong\u003e.\u003c/strong\u003e The moderating effect of sports-specialised customers on AI chatbot communication quality and flow experience\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFig\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003e4\u003c/strong\u003e\u003cstrong\u003e.\u0026nbsp;\u003c/strong\u003eThe moderating effect of sports-specialised customers on AI chatbot communication quality and perceived risk\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 4\u003c/strong\u003e. Regression Analysis Results (N=304)\u003c/p\u003e\n\u003cdiv\u003e\n \u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"677\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003ctd nowrap=\"\" style=\"width: 121px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" colspan=\"3\" style=\"width: 178px;\"\u003e\n \u003cp\u003eFlow Experience\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" colspan=\"3\" style=\"width: 192px;\"\u003e\n \u003cp\u003ePerceived Risk\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" colspan=\"3\" style=\"width: 185px;\"\u003e\n \u003cp\u003eImpulse Buying\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd nowrap=\"\" style=\"width: 121px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003eModel1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003eModel 2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003eModel 3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003eModel 4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003eModel 5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003eModel 6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003eModel 7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003eModel 8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 67px;\"\u003e\n \u003cp\u003eModel 9\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd nowrap=\"\" style=\"width: 121px;\"\u003e\n \u003cp\u003eGender\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.142*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.149**\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.143**\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e-0.032\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003e-0.038\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003e-0.031\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.106\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.068\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 67px;\"\u003e\n \u003cp\u003e0.097\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd nowrap=\"\" style=\"width: 121px;\"\u003e\n \u003cp\u003eAge\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.015\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.019\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.026\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e-0.032\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003e-0.036\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003e-0.044\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.053\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.049\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 67px;\"\u003e\n \u003cp\u003e0.043\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd nowrap=\"\" style=\"width: 121px;\"\u003e\n \u003cp\u003eYears of Online Shopping\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e-0.052\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e-0.054\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e-0.048\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.077\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003e0.079\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003e0.072\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e-0.027\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e-0.013\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 67px;\"\u003e\n \u003cp\u003e-0.003\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd nowrap=\"\" style=\"width: 121px;\"\u003e\n \u003cp\u003eOnline Shopping Frequency\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e-0.038\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e-0.025\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e-0.032\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e-0.008\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003e-0.019\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003e-0.011\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e-0.068\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e-0.057\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 67px;\"\u003e\n \u003cp\u003e-0.070\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd nowrap=\"\" style=\"width: 121px;\"\u003e\n \u003cp\u003eAI Chatbot Communication Quality\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.281***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.202***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.317***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.276***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003e0.342***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003e0.199*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.260***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.185***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 67px;\"\u003e\n \u003cp\u003e0.346***\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd nowrap=\"\" style=\"width: 121px;\"\u003e\n \u003cp\u003eFlow Experience\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.267***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 67px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd nowrap=\"\" style=\"width: 121px;\"\u003e\n \u003cp\u003ePerceived Risk\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 67px;\"\u003e\n \u003cp\u003e-0.310***\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd nowrap=\"\" style=\"width: 121px;\"\u003e\n \u003cp\u003eNon-Sports Specialised Consumers (Reference Group)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.273***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 67px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd nowrap=\"\" style=\"width: 121px;\"\u003e\n \u003cp\u003eSports-Specialized Consumers\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.276***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003e-0.229***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003e-0.233***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 67px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd nowrap=\"\" style=\"width: 121px;\"\u003e\n \u003cp\u003eAI Chatbot Communication Quality \u0026times; Sports-Specialised Consumers\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e-0.163*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003e0.201**\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 67px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd nowrap=\"\" style=\"width: 121px;\"\u003e\n \u003cp\u003eR \u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.096\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.164\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.176\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.082\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003e0.130\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003e0.149\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.080\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.144\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 67px;\"\u003e\n \u003cp\u003e0.168\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd nowrap=\"\" style=\"width: 121px;\"\u003e\n \u003cp\u003eAdjusted R\u003csup\u003e\u0026nbsp;2\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.080\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.147\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.157\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.066\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003e0.112\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003e0.129\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.064\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.127\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 67px;\"\u003e\n \u003cp\u003e0.151\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd nowrap=\"\" style=\"width: 121px;\"\u003e\n \u003cp\u003eF\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e6.304***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e9.678***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e9.055***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e5.296***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003e7.365***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 66px;\"\u003e\n \u003cp\u003e7.407***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e5.159***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 59px;\"\u003e\n \u003cp\u003e8.345***\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd nowrap=\"\" style=\"width: 67px;\"\u003e\n \u003cp\u003e9.970***\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003eNotes: *p \u0026lt; 0.05, **p \u0026lt; 0.01, ***p \u0026lt; 0.001\u003c/p\u003e"},{"header":"5. Discussion","content":"\u003cp\u003eThis study integrates the S-O-R framework with dual-process decision theory to reveal a dual-path mechanism through which AI chatbot communication quality influences impulse buying among sports-specialised customers. The research findings indicate that the communication quality of AI chatbots positively influenced impulse buying through flow experiences (H3) but negatively influenced impulse buying through perceived risk (H6). Specifically, the communication quality of AI chatbots significantly enhanced the flow experience (H1), which is consistent with the findings of Alalwan et al. (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2025\u003c/span\u003e) (Alalwan et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Flow experiences are characterised by immediate immersion and diminished self-awareness(Nakamura \u0026amp; Csikszentmihalyi, \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2009\u003c/span\u003e). When stimulated by the instant responses of AI chatbots, consumers experience significantly reduced waiting costs, making them more susceptible to being drawn into conversations and engaging with promotional discounts, product recommendations, and other information pushed by AI chatbots(Z. Cheng et al., \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). This immersive state weakens consumers\u0026rsquo; rational reflection ability, ultimately promoting impulse buying (H2). Moreover, the communication quality of AI chatbots positively influences perceived risk (H4). Perceived risk stems from an individual\u0026rsquo;s subjective assessment of potential losses(Dholakia, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2001\u003c/span\u003e; Xiaoxue Zhang \u0026amp; Yu, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2020\u003c/span\u003e), and its formation is closely linked to System 2 rational cognitive processing. Although AI chatbots offer advantages such as instant responses and personalised service(Song et al., \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2022\u003c/span\u003e), they may also trigger consumers\u0026rsquo; rational decision-making because of issues such as information specificity and demand matching. For sports-specialised customers, their specialised knowledge reserves trigger System 2 rational processing to scrutinise AI-generated information. Upon detecting discrepancies, they intensify perceived risk, which thereby suppresses impulse buying (H5). This finding aligns with Wu et al.\u0026rsquo;s (\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2020\u003c/span\u003e) finding that perceived risk negatively affects online impulse buying(I.-L. Wu et al., \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2020\u003c/span\u003e).\u003c/p\u003e\n\u003cp\u003eThis study further introduced sports-specialised customers as a moderating variable. The results indicated that sports-specialised customers both weakened the positive influence of AI chatbot communication quality on flow experience (H7) and strengthened its positive influence on perceived risk (H8). Moderation effect analysis revealed that the moderating effect of perceived risk among sports-specialised customers (\u0026beta;\u0026thinsp;=\u0026thinsp;0.201, p\u0026thinsp;\u0026lt;\u0026thinsp;0.01) was stronger than that of flow experience (\u0026beta; = -0.163, p\u0026thinsp;\u0026lt;\u0026thinsp;0.05). This difference stems from the fact that sports-specialised customers prioritise product performance and safety in their buying decisions(Chiu et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2018\u003c/span\u003e), with this objective taking precedence over the emotional pleasure elicited by flow experience. Consequently, when they interact with AI chatbots, sports-specialised customers are more inclined to allocate their expertise to risk identification. This goal-oriented cognitive resource allocation results in a more pronounced moderating effect along the perceived risk pathway, while the influence is relatively weaker along the flow experience pathway, which relies on intuition and emotion.\u003c/p\u003e\n\u003cdiv id=\"Sec23\" class=\"Section2\"\u003e\n \u003ch2\u003e5.1 Theoretical Contributions\u003c/h2\u003e\n \u003cp\u003eThis study makes two key theoretical contributions: (1) It pioneers an exploration of the mechanism through which AI (machine) influences customers\u0026rsquo; (human) online impulse buying, thereby expanding human‒computer interaction theory. On the basis of dual-system theory, this research reveals that System 2 cognitive processing heightens perceived risk during consumer interactions with AI chatbots, subsequently inhibiting impulse buying behaviour. This finding provides a novel perspective for understanding consumers\u0026rsquo; dual cognitive processing mechanisms in human‒computer interaction scenarios. Existing research has predominantly focused on the positive experiences derived from the communication characteristics of AI chatbots while paying insufficient attention to the potential cognitive conflicts they may induce in consumers(Alalwan et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Hao \u0026amp; Li, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). This study, which is grounded in dual-system theory, reveals the cognitive pathway through which AI chatbot communication quality influences purchase decisions. It finds that when interacting with AI chatbots, consumers simultaneously trigger System 1 intuitive processing because of communication fluency, leading to flow experiences, while also activating System 2 rational processing to evaluate the accuracy of AI-provided information, thereby heightening perceived risk. These findings corroborate prior research indicating that individuals are more receptive to AI recommendations in low-risk decision-making tasks(Longoni, Bonezzi, \u0026amp; Morewedge, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). However, in scenarios requiring expert knowledge assessment, AI recommendations may paradoxically diminish trust in AI judgments(Buechner et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2024\u003c/span\u003e), triggering scepticism about the information source. These findings not only clarify the rational decision-making pathway through which AI chatbot communication services influence consumer purchasing behaviour but also respond to scholars\u0026rsquo; calls for further research on how rational information affects impulse buying(Y. Wu et al., \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2021\u003c/span\u003e).\u003c/p\u003e\n \u003cp\u003e(2) By comparing professional judgment with AI precision recommendations, this study pioneers the comparison of human expertise versus AI expertise in new consumption scenarios. Examining the dual-system cognitive process of impulse buying among sports-specialised customers, this study explains the unique decision-making of professionals. Focusing on specific consumption scenarios and groups, it expands the boundaries of consumer behaviour research. Previous studies have often targeted general consumers and lacked targeted exploration of specific groups. In this study, the sports e-commerce context forms its entry point, and sports-specialised customers are treated as a moderating variable to explore differences in cognition and behaviour during interactions with AI chatbots. The findings indicate that the moderating effect of sports-specialised customers is significant across both pathways. On the one hand, the high cognitive standards formed by sports-specialised customers on the basis of their expertise diminish their perception of AI interaction fluency, thereby weakening the positive impact of AI chatbot communication quality on flow experiences. On the other hand, their expertise enhances their rational thinking ability, enabling them to evaluate AI-provided information and thereby amplifying the positive influence of communication quality on perceived risk. This result stems from the decision-making characteristics of expert consumers, who can focus precisely on useful information during product evaluation to form rational decisions(Redine et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Consequently, they can scrutinise product information and assess potential risks when they interact with AI chatbots. This study responds to scholars\u0026rsquo; calls for further exploration of how consumer expertise influences AI product recommendation effectiveness and aligns with Artem Redine et al.\u0026rsquo;s (\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2022\u003c/span\u003e) proposal to extend impulse-buying research into new contexts(Chinchanachokchai et al., \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Redine et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). By comparing expert judgment with AI precision recommendations, this study provides a model for applying relevant theories to specific populations and specialised domains, thereby enhancing theoretical explanatory power and universality.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec24\" class=\"Section2\"\u003e\n \u003ch2\u003e5.2 Practical Implications\u003c/h2\u003e\n \u003cp\u003eRelevant departments and industry associations should collaborate to establish information service standards for AI chatbots in specialised fields such as sporting goods. These standards should specify the information content that AI chatbots must disclose during product recommendations, thereby driving technological development aligned with industry needs. Concurrently, interdisciplinary communication platforms should be established to mitigate information biases in AI chatbot services within specialised domains, thereby fostering a regulated human‒machine interaction environment for consumer activities.\u003c/p\u003e\n \u003cp\u003eBusinesses can implement differentiated AI chatbot service strategies for distinct customer segments to meet their varied communication needs. For customers possessing specialised knowledge, emotional interactions should be minimised. On the one hand, embedding industry standards for products within the dialogue system enhances information accuracy. On the other hand, communication interfaces should be optimised to enable customers to independently access product-related inspection reports, thereby reducing perceived risk. For general consumers, the fluidity of AI chatbot interactions and personalised recommendations should be prioritised. Immersive interactions can promote consumers\u0026rsquo; flow experience, while appropriately integrating foundational knowledge dissemination balances emotional engagement with rational understanding.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec25\" class=\"Section2\"\u003e\n \u003ch2\u003e5.3 Limitations and Future Research Directions\u003c/h2\u003e\n \u003cp\u003eThis study focuses on the characteristics of sports-specialised customers as a customer group without further examining variations in their level of expertise or segmenting them on the basis of different types of sports. Customers engaged in different types of sports may have distinct professional demands for products. This singular approach to sample categorisation struggles to accurately capture customers\u0026rsquo; differentiated needs, hence limiting the generalizability of the research conclusions. Future research could enrich the dimensions of demographic analysis by incorporating factors such as expertise level, sport type, risk preference, and brand loyalty into existing classifications. This approach would allow for a deeper exploration of how these variables influence AI interaction decisions. Furthermore, this study focuses solely on text-based AI chatbots, whereas real-world consumer ecosystems feature diverse AI applications, such as VR/AR immersive shopping experiences, virtual influencer sales, and AI recommendation systems. Different AI interaction formats have distinct influence mechanisms on consumers. Future research could conduct cross-scenario comparative analyses to examine the differential effects of various AI technologies on flow experiences, perceived risk, and impulse buying behaviour, thereby clarifying the functional boundaries of different AI technologies. Furthermore, this study relied solely on two-stage questionnaire data for analysis. While this approach validated correlations between variables, it failed to dynamically capture the evolving cognitive and emotional states of consumers during AI interactions. Future research could integrate behavioural experiments by manipulating different dimensions of AI communication quality to identify the underlying mechanisms affecting consumer psychology and behaviour. Eye-tracking experiments could also be employed to capture consumers\u0026rsquo; emotional fluctuations and attention allocation during interactions, enabling deeper analysis of psychological mechanisms and enhancing the credibility of research conclusions.\u003c/p\u003e\n\u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eCompeting interests\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets generated during or analysed during the current study are available from the corresponding author on reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthical approval\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of Huaqiao University (Number: HQUIRB 202512\u0026ndash;11, Date: 26 December 2025). The scope of approval covered all research variables and data collection procedures in this study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eInformed consent\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study received ethical review approval on 26 December 2025, following which the electronic informed consent procedure was formally initiated. As this is a non-interventional online survey, the Institutional Review Board of Huaqiao University has granted an exemption from written informed consent. Our team obtain participants\u0026rsquo; electronic informed consent via an online questionnaire platform.\u0026nbsp;The period for collecting informed consent for the first round of questionnaires is from 27 December 2025 to 10 January 2026, following a 15-day interval, the period for collecting informed consent for the second round of questionnaires is from 26 January 2026 to 6 February 2026. The format of the electronic informed consent form is available on the questionnaire document. By clicking the informed consent confirmation button on the questionnaire\u0026rsquo;s home page, participants are deemed to have signed the informed consent form, agreeing to participate in this study and permitting the research team to collect and use the questionnaire data. All data will be anonymised and used solely for research analysis, report writing and the presentation of findings. The research team has fully informed participants of the study\u0026rsquo;s objectives, data usage protocols, anonymity safeguards and the absence of risks associated with participation. This study strictly adheres to the principle of voluntary participation; participants may refuse to participate or withdraw at any time without suffering any adverse consequences. This study does not involve vulnerable groups or minors.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor Contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eConceptualisation, Q.M. and Q.F.; Methodology, Q.M. and Q.F; Software, H.H.; Validation, W.Y., Q.Z. and H.H.; Formal Analysis, Q.F.; Investigation, Q.F., W.Y. and Q.Z.; Resources, Q.M. and H.H.; Data Curation, Q.F. and H.H.; Writing \u0026ndash; Original Draft Preparation, Q.F., W.Y. and Q.Z.; Writing \u0026ndash; Review \u0026amp; Editing, Q.M. and H.H.; Visualisation, Q.F. and W.Y.; Supervision, Q.M.; Project Administration, Q.M.; Funding Acquisition, Q.M. and H.H. All authors have read and agreed to the published version of the manuscript.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eAlalwan AA, Algharabat R, Abu El Samen A, Albanna H, Al-Okaily M (2025) Examining the impact of anthropomorphism and AI-chatbots service quality on online customer flow experience \u0026ndash; Exploring the moderating role of telepresence. J Consumer Mark 42(4):448\u0026ndash;471\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBarrouillet P (2011) Dual-process theories and cognitive development: Advances and challenges. Dev Rev 31(2):79\u0026ndash;85\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBeatty SE, Elizabeth Ferrell M (1998) Impulse buying: Modeling its precursors. J Retail 74(2):169\u0026ndash;191\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBuechner B, Stadler Blank A, Escoe B, Blaney I (2024) Expertise matters: Why consumers are less likely to follow AI recommendations in sports betting. Int J Sports Mark Spons 26(4):731\u0026ndash;747\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCabeza-Ram\u0026iacute;rez LJ, S\u0026aacute;nchez-Ca\u0026ntilde;izares SM, Santos-Rold\u0026aacute;n LM, Fuentes-Garc\u0026iacute;a FJ (2022) Impact of the perceived risk in influencers' product recommendations on their followers' purchase attitudes and intention. Technol Forecast Soc Chang 184:121997\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChang YP, Zhu DH (2012) The role of perceived social capital and flow experience in building users\u0026rsquo; continuance intention to social networking sites in China. Comput Hum Behav 28(3):995\u0026ndash;1001\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen C-C, Lin Y-C (2018) What drives live-stream usage intention? The perspectives of flow, entertainment, social interaction, and endorsement. Telematics Inform 35(1):293\u0026ndash;303\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen N, Yang Y (2023) The role of influencers in live streaming e-commerce: Influencer trust, attachment, and consumer purchase intention. J Theoretical Appl Electron Commer Res 18(3):1601\u0026ndash;1618\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen S, Li X, Liu K, Wang X (2023) Chatbot or human? The impact of online customer service on consumers' purchase intentions. Psychol Mark 40(11):2186\u0026ndash;2200\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCheng Y, Jiang H (2020) How do AI-driven chatbots impact user experience? Examining gratifications, perceived privacy risk, satisfaction, loyalty, and continued use. J Broadcast Electron Media 64(4):592\u0026ndash;614\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCheng Z, Fan W, Shao B, Jia W, Zhang Y (2024) The impact of intelligent customer service agents\u0026rsquo; initial response on consumers\u0026rsquo; continuous interaction intention. J Retailing Consumer Serv 76:103585\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChinchanachokchai S, Thontirawong P, Chinchanachokchai P (2021) A tale of two recommender systems: The moderating role of consumer expertise on artificial intelligence based product recommendations. J Retailing Consumer Serv 61:102528\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChiu W, Cho H, Chua H (2023) The dual roles of trust and risk in sport consumer decision-making in social commerce: An information adoption model. Sport Mark Q 32(4):267\u0026ndash;283\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChiu W, Kim T, Won D (2018) Predicting consumers\u0026rsquo; intention to purchase sporting goods online. Asia Pac J Mark Logistics 30(2):333\u0026ndash;351\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChung M, Ko E, Joung H, Kim SJ (2020) Chatbot e-service and customer satisfaction regarding luxury brands. J Bus Res 117:587\u0026ndash;595\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCoelho F, Aniceto I, Bairrada CM, Silva P (2023) Personal values and impulse buying: The mediating role of hedonic shopping motivations. J Retailing Consumer Serv 72:103236\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDalvi-Esfahani M, Barati-Ahmadabadi H, Ramayah T, Turner JJ, Iahad A, N., Azar N (2025) Stimulus-organism-response framework of decision-makers intention to adopt generative AI to replace entry-level jobs: The moderating impact of personality traits. Technol Forecast Soc Chang 219:124291\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDholakia UM (2001) A motivational process model of product involvement and consumer risk perception. Eur J Mark 35(11\u0026ndash;12):1340\u0026ndash;1362\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFeng Z, Al Mamun A, Masukujjaman M, Yang Q (2023) Modeling the significance of advertising values on online impulse buying behavior. Humanit Social Sci Commun 10(1):728\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGr\u0026uuml;nzner M, Richter I, White MP, Pahl S (2025) Hold on, do I really need this? Countering impulse clothing purchases through short reflections. Clean Responsible Consum 17:100280\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHao R, Li C (2025) How AI chatbots shape satisfactory experiences: A combined perspective of competence expansion and emotional extension. Technol Forecast Soc Chang 212:123979\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJacoby J (2002) Stimulus-organism-response reconsidered: An evolutionary step in modeling (consumer) behavior. J Consumer Psychol 12(1):51\u0026ndash;57\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKim J, Kim M, Lee S-M (2025) Unlocking trust dynamics: An exploration of playfulness, expertise, and consumer behavior in virtual influencer marketing. Int J Human\u0026ndash;Computer Interact 41(1):378\u0026ndash;390\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLee J-C, Xiong Ln (2024) The impact of informational antecedents on Generation Z consumers\u0026rsquo; evaluations of cross-border e-commerce transaction costs and purchase decisions. Technol Forecast Soc Chang 205:123513\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLee M, Park J-S (2022) Do parasocial relationships and the quality of communication with AI shopping chatbots determine middle-aged women consumers' continuance usage intentions? J Consumer Behav 21(4):842\u0026ndash;854\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLongoni C, Bonezzi A, Morewedge CK (2019) Resistance to medical artificial intelligence. J Consum Res 46(4):629\u0026ndash;650\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMilli S, Lieder F, Griffiths TL (2021) A rational reinterpretation of dual-process theories. Cognition 217:104881\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNakamura J, Csikszentmihalyi M (2009) Flow theory and research. Oxford University Press, New York, NY, US\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRedine A, Deshpande S, Jebarajakirthy C, Surachartkumtonkun J (2022) Impulse buying: A systematic literature review and future research directions. Int J Consumer Stud 47(1):3\u0026ndash;41\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRizomyliotis I, Kastanakis MN, Giovanis A, Konstantoulaki K, Kostopoulos I (2022) How mAy I help you today? The use of AI chatbots in small family businesses and the moderating role of customer affective commitment. J Bus Res 153:329\u0026ndash;340\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSengoz A, Cavusoglu M, Kement U, Bayar SB (2024) Unveiling the symphony of experience: Exploring flow, inspiration, and revisit intentions among music festival attendees within the SOR model. J Retailing Consumer Serv 81:104043\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShao B, Cheng Z, Wan L, Yue J (2021) The impact of cross border E-tailer's return policy on consumer's purchase intention. J Retailing Consumer Serv 59:102367\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShin H, Bunosso I, Levine LR (2022) The influence of chatbot humour on consumer evaluations of services. Int J Consumer Stud 47(2):545\u0026ndash;562\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSong M, Xing X, Duan Y, Cohen J, Mou J (2022) Will artificial intelligence replace human customer service? The impact of communication quality and privacy risks on adoption intention. J Retailing Consumer Serv 66:102900\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSpindler V, Schunk H, K\u0026ouml;necke T (2023) Sustainable consumption in sports fashion \u0026ndash; German runners' preference and willingness to pay for more sustainable sports apparel. Sustainable Prod Consum 42:411\u0026ndash;422\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTan R, Li Y, Huang Q, Liu H (2025) Enhancing customer service chatbot effectiveness: The effect of dyadic communication traits on customer purchase intention. J Association Inform Syst 26(3):799\u0026ndash;831\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTrivedi J, Kasilingam D, Arora P, Soni S (2022) The effect of augmented reality in mobile applications on consumers' online impulse purchase intention: The mediating role of perceived value. J Consumer Behav 21(4):896\u0026ndash;908\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUhm J-P, Kim S, Do C, Lee H-W (2022) How augmented reality (AR) experience affects purchase intention in sport E-commerce: Roles of perceived diagnosticity, psychological distance, and perceived risks. J Retailing Consumer Serv 67:103027\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWu I-L, Chiu M-L, Chen K-W (2020) Defining the determinants of online impulse buying through a shopping process of integrating perceived risk, expectation-confirmation model, and flow theory issues. Int J Inf Manag 52:102099\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWu Y, Xin L, Li D, Yu J, Guo J (2021) How does scarcity promotion lead to impulse purchase in the online market? A field experiment. Inf Manag 58(1):103283\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXin M, Jian L, Liu W, Bao Y (2025) Exploring the effect of live streaming atmospheric cues on consumer impulse buying: A flow experience perspective. J Theoretical Appl Electron Commer Res 20(2):149\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang X, Yu X (2020) The impact of perceived risk on consumers\u0026rsquo; cross-platform buying behavior. Front Psychol 11:592246\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang X, Zhang S (2024) Investigating impulse purchases in live streaming e-commerce: A perspective of match-ups. Technol Forecast Soc Chang 205:123513\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang Y, Ding Z, Sun J, Zhao X, Hu X, Yang Z (2025) Optimizing AI strategies in e-commerce customer service: An agent-based simulation. Electron Markets 35(1):77\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"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":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"AI chatbots, Impulsive buying behaviour, Flow experience, Perceived risk, Sports-specialised consumers","lastPublishedDoi":"10.21203/rs.3.rs-9008587/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9008587/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eIn the context of human\u0026ndash;AI collaborative consumption, AI chatbots have emerged as a critical tool for enhancing consumer conversion rates on e-commerce platforms. Their influence on impulsive buying behaviour among general consumers has garnered significant academic and industry attention. However, the extent to which AI-driven recommendation systems may influence expert judgment requires further empirical investigation. To investigate the impact of interactions between sports-specialised consumers and AI chatbots on impulsive buying behaviour with respect to athletic training products, this study integrates the S-O-R model with dual-systems theory and constructs a dual-path cognitive mechanism. It explores the mediating roles of flow experience and perceived risk, as well as the moderating effects of sports-specialised consumers, within the dual-path framework. The results indicate that AI service quality positively influences the impulsive buying tendencies of sports-specialised consumers through flow experience, that AI service quality negatively affects their impulsive buying tendencies through risk perception, and that customers with sports-related expertise serve as a moderating variable, attenuating the positive effect of flow experience induced by AI chatbots and amplifying the positive influence of perceived risk. This study elucidates the cognitive processing pathways of sports-specialised consumers in human\u0026ndash;computer interactions and provides empirical evidence to assist e-commerce enterprises in optimising AI service design for targeted demographic segments.\u003c/p\u003e","manuscriptTitle":"Are Professionals also influenced by AI to Make Impulse Buying? A Dual-System Cognitive Process of Sports-Specialised Customers","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-15 11:06:23","doi":"10.21203/rs.3.rs-9008587/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"ea01d8eb-6db8-4a19-8915-ef9dc9c31294","owner":[],"postedDate":"April 15th, 2026","published":true,"recentEditorialEvents":[{"type":"decision","content":"Rejected","date":"2026-04-29T21:33:30+00:00","index":"","fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":66052719,"name":"Business and commerce/Business and management"},{"id":66052720,"name":"Social science/Business and management"},{"id":66052721,"name":"Business and commerce/Information systems and information technology"},{"id":66052722,"name":"Biological sciences/Psychology"},{"id":66052723,"name":"Social science/Psychology"},{"id":66052724,"name":"Social science/Science technology and society"}],"tags":[],"updatedAt":"2026-04-29T21:39:44+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-15 11:06:23","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9008587","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9008587","identity":"rs-9008587","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}