Artificial intelligence drives divergent emission futures | 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 Artificial intelligence drives divergent emission futures Hamish Beath, Shivika Mittal, Ajay Gambhir, Samira Barzin, John Bistline, and 6 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8582166/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted You are reading this latest preprint version Abstract Artificial Intelligence (AI) may impact greenhouse gas emissions in both positive and negative ways, yet these impacts are not represented in recent assessments of climate change mitigation. Uncertainty surrounding future AI capability and adoption, its impacts on energy systems, and limited data availability complicates the inclusion of AI impacts. We conduct an expert-led process, ascertaining plausible impacts, and quantify ranges in their magnitude to 2040. We present a new scenario framework for AI and climate change that considers interactions between AI growth, climate policy, and AI policy. We translate elicited insights into illustrative scenarios and assess their implications using an integrated assessment model. Cumulative global CO2 emissions by 2040 range from +11% above to 1.4% below a baseline without AI impacts, depending on the scale of AI growth and policy intervention. Our results highlight emissions risks and opportunities from AI and the need for explicit representation in mitigation scenarios. Earth and environmental sciences/Environmental sciences/Environmental impact Scientific community and society/Energy and society/Energy supply and demand Scientific community and society/Social sciences/Climate change/Climate-change mitigation Full Text Additional Declarations There is NO Competing Interest. Ethics approval: The study involved expert elicitation of consenting professionals. No sensitive personal data was collected, participation was voluntary, and responses were anonymised. Ethics approval from Imperial College was therefore not sought.Participant consent: All expert participants were given full information regarding the study design and data that would be collected. All experts provided full informed consent to participate. Supplementary Files SupplementaryInformationAIpaper.pdf Supplementary Information Cite Share Download PDF Status: Under Review 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-8582166","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":580606653,"identity":"e43da306-7abd-41b0-a395-a087c7047cdc","order_by":0,"name":"Hamish Beath","email":"data:image/png;base64,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","orcid":"https://orcid.org/0000-0002-5124-9143","institution":"Imperial College London","correspondingAuthor":true,"prefix":"","firstName":"Hamish","middleName":"","lastName":"Beath","suffix":""},{"id":580606654,"identity":"f3028d66-c23b-471d-b7fe-28e701fdda20","order_by":1,"name":"Shivika Mittal","email":"","orcid":"https://orcid.org/0000-0003-4718-0064","institution":"CICERO Center for International Climate Research","correspondingAuthor":false,"prefix":"","firstName":"Shivika","middleName":"","lastName":"Mittal","suffix":""},{"id":580606655,"identity":"9c85f0ae-ec35-4cc2-bc42-0d731ab4b6dc","order_by":2,"name":"Ajay Gambhir","email":"","orcid":"","institution":"Accelerator for Systemic Risk Assessment, UN Foundation, Washington DC, USA","correspondingAuthor":false,"prefix":"","firstName":"Ajay","middleName":"","lastName":"Gambhir","suffix":""},{"id":580606656,"identity":"f3643bbf-64da-4ac0-b34d-069b3f462c68","order_by":3,"name":"Samira Barzin","email":"","orcid":"","institution":"University of Oxford, Environmental Change Institute","correspondingAuthor":false,"prefix":"","firstName":"Samira","middleName":"","lastName":"Barzin","suffix":""},{"id":580606657,"identity":"8bf03912-497d-48e0-8ed5-62d26e335b66","order_by":4,"name":"John Bistline","email":"","orcid":"","institution":"EPRI","correspondingAuthor":false,"prefix":"","firstName":"John","middleName":"","lastName":"Bistline","suffix":""},{"id":580606658,"identity":"9b4978bd-6b93-485d-9410-14ed93acf718","order_by":5,"name":"Felix Creutzig","email":"","orcid":"https://orcid.org/0000-0002-5710-3348","institution":"Potsdam Institute for Climate Impact Research","correspondingAuthor":false,"prefix":"","firstName":"Felix","middleName":"","lastName":"Creutzig","suffix":""},{"id":580606659,"identity":"90005976-853d-44fd-a715-d418ce5f79f7","order_by":6,"name":"Lynn Kaack","email":"","orcid":"https://orcid.org/0000-0003-3630-3102","institution":"Hertie School","correspondingAuthor":false,"prefix":"","firstName":"Lynn","middleName":"","lastName":"Kaack","suffix":""},{"id":580606660,"identity":"ade35c42-efde-40da-873b-076aade9d1ce","order_by":7,"name":"George Kamiya","email":"","orcid":"https://orcid.org/0000-0003-4376-0790","institution":"Reuben College, University of Oxford, Oxford, United Kingdom","correspondingAuthor":false,"prefix":"","firstName":"George","middleName":"","lastName":"Kamiya","suffix":""},{"id":580606661,"identity":"1458267b-7cc6-4d81-ae4e-ddf9e2d38ff1","order_by":8,"name":"Charlie Wilson","email":"","orcid":"https://orcid.org/0000-0001-8164-3566","institution":"Environmental Change Institute, University of Oxford","correspondingAuthor":false,"prefix":"","firstName":"Charlie","middleName":"","lastName":"Wilson","suffix":""},{"id":580606662,"identity":"c1276db4-2c83-4ad6-8f2b-ccc77f0bf296","order_by":9,"name":"Joeri Rogelj","email":"","orcid":"https://orcid.org/0000-0003-2056-9061","institution":"Imperial College London","correspondingAuthor":false,"prefix":"","firstName":"Joeri","middleName":"","lastName":"Rogelj","suffix":""},{"id":580606663,"identity":"dc3111bc-7cff-4a8d-8d66-bb51dc5c4b59","order_by":10,"name":"Robin Lamboll","email":"","orcid":"https://orcid.org/0000-0002-8410-037X","institution":"Imperial College","correspondingAuthor":false,"prefix":"","firstName":"Robin","middleName":"","lastName":"Lamboll","suffix":""}],"badges":[],"createdAt":"2026-01-12 13:15:19","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8582166/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8582166/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":103508102,"identity":"c4c1a8a6-39ed-4d9c-b2c9-83460f6cfa35","added_by":"auto","created_at":"2026-02-26 13:47:13","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":978174,"visible":true,"origin":"","legend":"Article File","description":"","filename":"AIPaperSubmissionDraft.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8582166/v1_covered_24669d97-6d16-4a50-9b39-1ac8d1d0e5f1.pdf"},{"id":103208634,"identity":"c4d35cfa-5eba-451f-bd61-1cd98404899d","added_by":"auto","created_at":"2026-02-23 08:03:41","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":178775,"visible":true,"origin":"","legend":"Supplementary Information","description":"","filename":"SupplementaryInformationAIpaper.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8582166/v1/a0cb665987ae7f5c245a401b.pdf"}],"financialInterests":"\u003cp\u003eThere is \u003cstrong\u003eNO\u003c/strong\u003e Competing Interest.\u003c/p\u003e\n\u003cp\u003eEthics approval: The study involved expert elicitation of consenting professionals. No sensitive personal data was collected, participation was voluntary, and responses were anonymised. Ethics approval from Imperial College was therefore not sought.Participant consent: All expert participants were given full information regarding the study design and data that would be collected. All experts provided full informed consent to participate.\u003c/p\u003e","formattedTitle":"Artificial intelligence drives divergent emission futures","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"
[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-8582166/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8582166/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Artificial Intelligence (AI) may impact greenhouse gas emissions in both positive and negative ways, yet these impacts are not represented in recent assessments of climate change mitigation. Uncertainty surrounding future AI capability and adoption, its impacts on energy systems, and limited data availability complicates the inclusion of AI impacts. We conduct an expert-led process, ascertaining plausible impacts, and quantify ranges in their magnitude to 2040. We present a new scenario framework for AI and climate change that considers interactions between AI growth, climate policy, and AI policy. We translate elicited insights into illustrative scenarios and assess their implications using an integrated assessment model. Cumulative global CO2 emissions by 2040 range from +11% above to 1.4% below a baseline without AI impacts, depending on the scale of AI growth and policy intervention. Our results highlight emissions risks and opportunities from AI and the need for explicit representation in mitigation scenarios.","manuscriptTitle":"Artificial intelligence drives divergent emission futures","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-02-23 08:03:31","doi":"10.21203/rs.3.rs-8582166/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"
[email protected]","identity":"nature-climate-change","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"nclimate","sideBox":"Learn more about [Nature Climate Change](http://www.nature.com/nclimate/)","snPcode":"","submissionUrl":"","title":"Nature Climate Change","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Nature Research","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"a6c578f6-90d6-402c-8756-c18f40d66db7","owner":[],"postedDate":"February 23rd, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":61765070,"name":"Earth and environmental sciences/Environmental sciences/Environmental impact"},{"id":61765071,"name":"Scientific community and society/Energy and society/Energy supply and demand"},{"id":61765072,"name":"Scientific community and society/Social sciences/Climate change/Climate-change mitigation"}],"tags":[],"updatedAt":"2026-03-02T12:50:15+00:00","versionOfRecord":[],"versionCreatedAt":"2026-02-23 08:03:31","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8582166","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8582166","identity":"rs-8582166","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}
Text is read by the "Ask this paper" AI Q&A widget below.
Extraction quality varies by source — PMC NXML preserves structure
cleanly, OA-HTML may include some navigation residue, and OA-PDF can
have broken hyphenation. The publisher copy
(via DOI)
is the canonical version.