{"id":1151483,"date":"2025-10-07T15:09:32","date_gmt":"2025-10-07T22:09:32","guid":{"rendered":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/?post_type=msr-research-item&p=1151483"},"modified":"2025-12-16T14:10:07","modified_gmt":"2025-12-16T22:10:07","slug":"learning-outcomes-with-genai-in-the-classroom-a-review-of-empirical-evidence","status":"publish","type":"msr-research-item","link":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/publication\/learning-outcomes-with-genai-in-the-classroom-a-review-of-empirical-evidence\/","title":{"rendered":"Learning outcomes with GenAI in the classroom: A review of empirical evidence"},"content":{"rendered":"

This report presents a review of recent empirical evidence of generative AI (GenAI) impact on learning outcomes in formal education. Its purpose is to provide educators with an overview of top concerns for ensuring students\u2019 learning gains when using LLM-based learning tools and concludes with research-derived guidance for deciding when and how to use these tools in the classroom. The report\u00a0unfolds\u00a0as follows:\u00a0<\/span>\u00a0<\/span><\/p>\n

Section 1 distinguishes between the needs of education and industry<\/strong>,<\/span>\u00a0where the benefits of LLMs were first explored, primarily for productivity gains.\u00a0Educators\u2019 priorities are different.\u00a0Pedagogical concerns include\u00a0consideration of\u00a0inequities in education, developing students\u2019 critical thinking skills, and the potential for\u00a0GenAI\u00a0to inhibit social development. These concerns extend beyond technologists\u2019 focus on mitigating technical harms such as toxic content, bias, or accuracy in system outputs.\u00a0<\/span>\u00a0<\/span><\/p>\n

Section 2 presents several key variables\u00a0that affect learning\u00a0with\u00a0GenAI:<\/strong>\u00a0<\/span>(1)<\/span>\u00a0AI literacy\u2014understanding the capabilities and limitations of an AI system\u2014is a critical new variable for student success when using\u00a0GenAI.\u00a0<\/span>(2)\u00a0<\/span>Educational equity is a variable where\u00a0GenAI\u00a0renders mixed experiences for marginalized groups. Studies show how\u00a0GenAI\u00a0can be an effective resource for students with disabilities. In other contexts, it entrenches existing patterns in academic performance of the weakest students and can\u00a0exacerbate\u00a0inequities for economically marginalized students.\u00a0<\/span>(3)<\/span>\u00a0GenAI\u00a0can\u00a0impact\u00a0psychological and social conditions long recognized to\u00a0facilitate\u00a0learning: self-efficacy, individual pace, and human connection. On self-efficacy, studies show that students can be overconfident about their skill mastery when using\u00a0GenAI\u00a0and need help calibrating their mental model of learning gains. For self-paced learning,\u00a0GenAI\u00a0introduces both efficiencies and pitfalls depending on\u00a0learning domain and\u00a0context, including whether AI tools are general purpose chatbots or scaffolded tutors.\u00a0Studies also highlight\u00a0GenAI\u00a0impact on human connection, the foundation for developing higher-order skills of critical thinking and creativity.\u00a0GenAI\u2019s\u00a0on-demand availability but lack of social presence\u00a0can present opportunities and disadvantages, from providing a nonjudgmental environment for exploring topics to reducing collaboration with peers in group projects. Yet, studies show that human tutors\u00a0remain\u00a0students\u2019 preferred source for trusted information.\u00a0<\/span>\u00a0<\/span><\/p>\n

Section 3 examines how\u00a0GenAI\u00a0usage aligns with learning\u00a0objectives\u00a0in Bloom\u2019s taxonomy.<\/strong>\u00a0<\/span><\/b>Basic cognitive skills\u2014Bloom\u2019s\u00a0<\/span>remembering<\/span><\/i>\u00a0and\u00a0<\/span>understanding<\/span><\/i>\u2014are fundamental to success across academic domains. Studies show that there can be\u00a0an overdependence\u00a0and lack of engagement that\u00a0result\u00a0in impaired memory\u00a0formation when using LLM chatbots. Development of higher-order thinking\u2014<\/span>analysis, reasoning,<\/span><\/i>\u00a0and\u00a0<\/span>creativity<\/span><\/i>\u2014can be compromised if\u00a0GenAI\u00a0is used in ways that bypass the necessary struggle that is integral to\u00a0acquiring\u00a0skills. Studies illustrate how use of general-purpose\u00a0GenAI\u00a0tools such as ChatGPT, without scaffolding or other pedagogical guardrails, can be detrimental to critical thinking.\u00a0GenAI\u00a0can also\u00a0impact\u00a0creativity. Students using\u00a0GenAI\u00a0for creative problem-solving can\u00a0benefit\u00a0from fast prototype iteration and greater project completeness or detail but can also tend toward idea fixation and less originality and complexity in their work.<\/span>\u00a0<\/span><\/p>\n

Section 4 highlights how\u00a0GenAI\u00a0learning tools need greater pedagogical complexity<\/strong>.<\/span>\u00a0Up to now,\u00a0state-of-the-art\u00a0tools have been ChatGPT or similar, with prompt engineering for the model to assume an instructor role or restrain its\u00a0outputs. However,\u00a0modified\u00a0general-purpose chatbots cannot address the broad range of pedagogical considerations involved in learning success. New types of experimental AI tutors with embedded proven pedagogical strategies\u2014for example, capable of detecting and effectively responding to a range of student cognitive states\u2014show promise. Consulting educators in\u00a0the design\u00a0is key for success of systems like these that are on the horizon.<\/span>\u00a0<\/span><\/p>\n

A concluding synthesis of the empirical evidence offers four guidelines for integrating\u00a0GenAI\u00a0in learning environments:<\/strong> (1)<\/span>\u00a0<\/span><\/b>Ensure student readiness\u2014avoid introducing\u00a0GenAI\u00a0too early, before students master domain basics.\u00a0<\/span>(2)<\/span>\u00a0<\/span><\/b>Teach AI literacy\u2014build an awareness of\u00a0GenAI\u00a0capabilities and limitations so students can assess system outputs and learn domain-specific techniques for\u00a0optimal\u00a0results.\u00a0<\/span>(3)<\/span>\u00a0<\/span><\/b>Use\u00a0GenAI\u00a0as a supplement to traditional learning methods\u2014GenAI\u00a0explanations and examples are capabilities that students value, but teacher guidance with these explanations\u00a0remains\u00a0necessary.\u00a0<\/span>(4)<\/span>\u00a0<\/span><\/b>Promote design interventions that foster student\u00a0engagement\u2014limiting copy-paste functionality, supporting students\u2019 metacognitive calibration to reduce overestimation of their learning progress, nudging learners\u00a0towards\u00a0critical thinking, and evaluating\u00a0GenAI\u00a0tools for proven\u00a0engagement strategies.<\/span>\u00a0<\/span><\/p>\n

 <\/p>\n

Cite as:
\nWalker, K. and Vorvoreanu, M. 2025.<\/strong> Learning outcomes with GenAI in the classroom: A review of empirical evidence.<\/em> Microsoft Technical Report MSR-TR-2025-42 October 2025.<\/p>\n","protected":false},"excerpt":{"rendered":"

This report presents a review of recent empirical evidence of generative AI (GenAI) impact on learning outcomes in formal education. Its purpose is to provide educators with an overview of top concerns for ensuring students\u2019 learning gains when using LLM-based learning tools and concludes with research-derived guidance for deciding when and how to use these […]<\/p>\n","protected":false},"featured_media":0,"template":"","meta":{"msr-url-field":"","msr-podcast-episode":"","msrModifiedDate":"","msrModifiedDateEnabled":false,"ep_exclude_from_search":false,"_classifai_error":"","msr-author-ordering":null,"msr_publishername":"","msr_publisher_other":"","msr_booktitle":"","msr_chapter":"","msr_edition":"","msr_editors":"","msr_how_published":"","msr_isbn":"","msr_issue":"","msr_journal":"","msr_number":"MSR-TR-2025-42","msr_organization":"Microsoft","msr_pages_string":"","msr_page_range_start":"","msr_page_range_end":"","msr_series":"","msr_volume":"","msr_copyright":"","msr_conference_name":"","msr_doi":"","msr_arxiv_id":"","msr_s2_paper_id":"","msr_mag_id":"","msr_pubmed_id":"","msr_other_authors":"","msr_other_contributors":"","msr_speaker":"","msr_award":"","msr_affiliation":"","msr_institution":"","msr_host":"","msr_version":"","msr_duration":"","msr_original_fields_of_study":"","msr_release_tracker_id":"","msr_s2_match_type":"","msr_citation_count_updated":"","msr_published_date":"2025-10-1","msr_highlight_text":"","msr_notes":"","msr_longbiography":"","msr_publicationurl":"","msr_external_url":"","msr_secondary_video_url":"","msr_conference_url":"","msr_journal_url":"","msr_s2_pdf_url":"","msr_year":0,"msr_citation_count":0,"msr_influential_citations":0,"msr_reference_count":0,"msr_s2_match_confidence":0,"msr_microsoftintellectualproperty":true,"msr_s2_open_access":false,"msr_s2_author_ids":[],"msr_pub_ids":[],"msr_hide_image_in_river":null,"footnotes":""},"msr-research-highlight":[],"research-area":[13556],"msr-publication-type":[193718],"msr-publisher":[],"msr-focus-area":[],"msr-locale":[268875],"msr-post-option":[269148,269142],"msr-field-of-study":[],"msr-conference":[],"msr-journal":[],"msr-impact-theme":[],"msr-pillar":[],"class_list":["post-1151483","msr-research-item","type-msr-research-item","status-publish","hentry","msr-research-area-artificial-intelligence","msr-locale-en_us","msr-post-option-approved-for-river","msr-post-option-include-in-river"],"msr_publishername":"","msr_edition":"","msr_affiliation":"","msr_published_date":"2025-10-1","msr_host":"","msr_duration":"","msr_version":"","msr_speaker":"","msr_other_contributors":"","msr_booktitle":"","msr_pages_string":"","msr_chapter":"","msr_isbn":"","msr_journal":"","msr_volume":"","msr_number":"MSR-TR-2025-42","msr_editors":"","msr_series":"","msr_issue":"","msr_organization":"Microsoft","msr_how_published":"","msr_notes":"","msr_highlight_text":"","msr_release_tracker_id":"","msr_original_fields_of_study":"","msr_download_urls":"","msr_external_url":"","msr_secondary_video_url":"","msr_longbiography":"","msr_microsoftintellectualproperty":1,"msr_main_download":"","msr_publicationurl":"","msr_doi":"","msr_publication_uploader":[{"type":"file","viewUrl":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-content\/uploads\/2025\/10\/GenAILearningOutcomes_published_2025-12-16.pdf","id":"1158762","title":"genailearningoutcomes_published_2025-12-16","label_id":"243109","label":0}],"msr_related_uploader":"","msr_citation_count":0,"msr_citation_count_updated":"","msr_s2_paper_id":"","msr_influential_citations":0,"msr_reference_count":0,"msr_arxiv_id":"","msr_s2_author_ids":[],"msr_s2_open_access":false,"msr_s2_pdf_url":null,"msr_attachments":[{"id":1158762,"url":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-content\/uploads\/2025\/10\/GenAILearningOutcomes_published_2025-12-16.pdf"},{"id":1151484,"url":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-content\/uploads\/2025\/10\/GenAILearningOutcomes-Report-published-10-07-2025.pdf"}],"msr-author-ordering":[{"type":"text","value":"Kathy Walker","user_id":0,"rest_url":false},{"type":"user_nicename","value":"Mihaela Vorvoreanu","user_id":36804,"rest_url":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Mihaela Vorvoreanu"}],"msr_impact_theme":[],"msr_research_lab":[],"msr_event":[],"msr_group":[],"msr_project":[1151473],"publication":[],"video":[],"msr-tool":[],"msr_publication_type":"techreport","related_content":{"projects":[{"ID":1151473,"post_title":"Psychological influences of AI","post_name":"psychological-influences-of-ai","post_type":"msr-project","post_date":"2025-10-07 13:27:35","post_modified":"2026-05-13 13:34:19","post_status":"publish","permalink":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/project\/psychological-influences-of-ai\/","post_excerpt":"The Psychological Influences of AI (Psi) project explores how AI\u2014especially generative AI\u2014affects human psychology, including cognition, emotion, behavior, and well-being. It investigates topics such as trust, dependency, attention, agency, the psychological impacts of anthropomorphic AI, people forming problematic relationship with AI, and AI-associated delusions. One example of a project we are actively working on is understanding, through qualitative in-depth research interviews, those moments when people felt maybe AI was too real, understood or was aware…","_links":{"self":[{"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-project\/1151473"}]}}]},"_links":{"self":[{"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/1151483","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item"}],"about":[{"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/types\/msr-research-item"}],"version-history":[{"count":2,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/1151483\/revisions"}],"predecessor-version":[{"id":1151486,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/1151483\/revisions\/1151486"}],"wp:attachment":[{"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/media?parent=1151483"}],"wp:term":[{"taxonomy":"msr-research-highlight","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-research-highlight?post=1151483"},{"taxonomy":"msr-research-area","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/research-area?post=1151483"},{"taxonomy":"msr-publication-type","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-publication-type?post=1151483"},{"taxonomy":"msr-publisher","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-publisher?post=1151483"},{"taxonomy":"msr-focus-area","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-focus-area?post=1151483"},{"taxonomy":"msr-locale","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-locale?post=1151483"},{"taxonomy":"msr-post-option","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-post-option?post=1151483"},{"taxonomy":"msr-field-of-study","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-field-of-study?post=1151483"},{"taxonomy":"msr-conference","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-conference?post=1151483"},{"taxonomy":"msr-journal","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-journal?post=1151483"},{"taxonomy":"msr-impact-theme","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-impact-theme?post=1151483"},{"taxonomy":"msr-pillar","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-pillar?post=1151483"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}