MIT committee reports AI can produce credible solutions to almost any undergraduate assignment
MIT released its Ad Hoc Committee report on 25 August 2026, finding AI can answer almost any undergraduate assignment and urging every course to state a policy.
Photo by Gera Cejas on Pexels
The Massachusetts Institute of Technology published the report of its Ad Hoc Committee on AI Use in Teaching, Learning, and Research Training on 25 August 2026, and the MIT AI report states that generative AI “can produce credible solutions and provide reasonable responses to almost any written assignment in our undergraduate curriculum, including essays, math and science problems, proofs, and coding assignments, and their power will only grow”. The document is dated 13 August 2026 and runs to 38 pages with four appendices. It recommends that every MIT course adopt and state an explicit AI use policy, that the institute invest more rather than less in its residential community, and that MIT build permanent machinery for testing and revising its approach.
Chancellor Melissa Nobles, Provost Anantha Chandrakasan and Chair of the Faculty Roger Levy charged the committee on 14 January 2026 with three tasks: “Assess current AI use at MIT”, “Identify innovations in teaching and student assessment” and “Propose an AI use policy”. The committee met weekly through the spring, ran surveys, reviewed peer institutions’ policies and held listening sessions. It was co-chaired by Eric Klopfer, professor in Comparative Media Studies/Writing, and Sam Madden, Distinguished College of Computing Professor in Electrical Engineering and Computer Science, with 16 further members including Libraries director Chris Bourg, Media Arts and Sciences professor Cynthia Breazeal and two undergraduates. The committee also disclosed its own AI use: no report text was AI-generated, though early drafts were passed to ChatGPT to find overlapping sections and some Appendix C graphs were produced with OpenAI’s Codex.
What the MIT AI report measured
The report draws on three separate surveys, and the numbers should not be blended. The committee ran its own AI Usage and Attitudes Survey in spring 2026, added AI questions to MIT’s spring 2026 Quality of Life Survey, and summarised a Fall 2025 survey by the student newspaper The Tech.
| Survey | Population | Responses | Response rate |
|---|---|---|---|
| MIT AI Usage and Attitudes Survey, spring 2026 | Faculty, instructional staff and students | 1,632 | 12% |
| MIT Quality of Life Survey, spring 2026 | All main-campus community members | about 8,200 | Not stated |
| The Tech LLM survey, fall 2025 | 659 undergraduates, 248 graduate students and postdocs, 18 faculty, 72 staff | 1,002 | Not stated |
Source: MIT Ad Hoc Committee report, Appendix C, 13 August 2026.
On the committee’s own survey, 44 per cent of all respondents said they use ChatGPT often or very often, split 46 per cent of students and 35 per cent of instructors. The report notes that Sloan “reported a much higher rate of usage of all tools than any other group”, and that school-level response rates ranged from 20 per cent in SHASS to 10 per cent at Sloan.
The Quality of Life Survey produced the attitude figures. About 40 per cent of respondents said they used AI tools often or very often; 53 per cent of service and support staff said they never used them. Only 23 per cent rated themselves optimistic about generative AI, scoring a 4 or 5 on a pessimism-to-optimism scale.
One line in Appendix C reads “46% of undergrads and 60% of undergrads and graduate students reporting at least often use”. The same wording appears in both the PDF and the web appendices, so it is the report’s own text; the second figure appears to be intended for graduate students, but the report does not say so.
The Tech’s fall 2025 survey supplies the sharpest teaching numbers. It found 46 per cent of undergraduates using large language models daily and 30 per cent several times per week, with over 80 per cent using them to have course material explained and around 70 per cent for coding help. It also found 90 per cent of undergraduates somewhat or very concerned about overreliance, including 67 per cent very concerned. Appendix C summarises the preparation gap in one line: “Although 70% agreed that AI proficiency will be important in their careers, only 25% believed MIT is preparing students to use AI professionally.”
| Task | Undergraduates | Graduate students and postdocs |
|---|---|---|
| Programming or coding assistance | around 70% | 88% |
| Explaining course material | over 80% | Not reported |
| Completing coursework | over 50% | Not reported |
| Summarising papers | over 50% | 52% |
| Brainstorming ideas | Not reported | 55% |
| Writing essays | Not reported | 48% |
Source: The Tech, LLM survey results, 25 November 2025, as summarised in MIT report Appendix C.3.
What the committee says has changed on campus
The report attributes a set of behavioural shifts to AI adoption without attaching numbers to any of them. It states that “in less than three years these technologies have driven major shifts in campus culture, including decreased attendance at office hours, reduced participation in online discussions, and, as we heard anecdotally, a drop in in-person study groups in dorms, libraries, and other study spaces”. The study-group claim is explicitly flagged as anecdotal. Elsewhere the committee lists AI as eroding the “social contract” between instructors and students and making student progress harder to assess. On learning itself, the report warns that “getting the right answer from a chatbot can create the illusion of learning”.
The report also records a student objection the committee heard about instructors’ own AI use: “Why should I bother coming to class or doing the work if the teacher is just going to give an AI-generated lecture?” One recommendation asks instructors to disclose their own AI use for that reason.
The 25 sub-recommendations sit under three headings: adapt educational processes for an AI-aware world; centre people, community and the residential experience; and build processes, teams and tools for continuous reflection, iteration and improvement. Eight guiding principles precede them, including “Be humble”, “Augmentation not automation” and “No one size fits all”.
Appendix B sets out a menu of four course policies that instructors are asked to state on syllabi, with traffic-light icons to make the level legible at a glance.
| Policy | Wording in the report | Committee note |
|---|---|---|
| Unrestricted GenAI use | “Students may use any GenAI system for any purpose on assignments in this course.” | – |
| Limited GenAI use: support tool only | “Students may use GenAI to support learning, brainstorming, editing, debugging, or generating explanations, but may not use it to produce full or substantial assignment solutions.” | “We expect that this is the appropriate policy for many of our courses.” |
| Required GenAI use | “Students are required to use GenAI to complete the assignment in the manner specified.” | – |
| GenAI use strictly prohibited | “GenAI may not be used in any form for the course.” | – |
Source: MIT Ad Hoc Committee report, Appendix B, 13 August 2026.
On enforcement, the committee recommends against relying on AI detectors, “as it risks an arms race”, and notes that MIT’s Committee on Discipline “does not consider AI detector output alone sufficient”. It describes the current generation of lockdown browsers as “buggy, error-prone” and says it “feels like surveillance”. Instead it points to oral exams, semester portfolios, and out-of-class assignments paired with in-class conversations. It explicitly rejects grade rationing, stating that it does not advocate “imposing a limit on the number of A’s that can be given”. On research writing, the rule is short: “AI should never be listed as a co-author”, and all theses should carry a statement of how AI was used.
Structural recommendations include an ongoing AI and education committee, AI Leads at school or department level, funded AI Fellows and an implementation team, an AI Pilot Fund, and policies covering logging, auditing, data sensitivity, equitable access and the environmental cost of AI. The report also documents MIT’s own platform, Parley, which is model-agnostic and gives each user up to $30 a month in free credits, against top commercial plans costing as much as $200 a month as of June 2026.
President Sally Kornbluth released the report with a letter the same day. She wrote that “AI presents itself to us as a kind of superpower, with both extraordinary potential and disturbing risks”, and that “far and away the most important work will be cultural”. The covering letter from Nobles, Chandrakasan and Levy called it “a strong call to action”. The report opens with the same phrase: “This report is a call to action.”
The findings extend a pattern Winss Solutions has tracked across universities this year, including Brown’s first generative AI in teaching report, Purdue making AI coursework a graduation requirement, and Turnitin’s finding that US students lead peers in AI writing use. MIT’s caution on detection tools matches the pattern described in coverage of AI in higher education reaching a tipping point.
About MIT’s work on AI and education
MIT’s institutional work on AI in education predates the committee. In May 2021 the institute launched RAISE, short for Responsible AI for Social Empowerment and Education, headquartered in the MIT Media Lab as a collaboration with the MIT Schwarzman College of Computing and MIT Open Learning. Cynthia Breazeal directs it; Eric Klopfer was one of its co-directors at launch, alongside Hal Abelson and Hae Won Park. RAISE started the Day of AI programme in 2021, offering free K-12 AI curriculum, which later became a stand-alone nonprofit. MIT’s Social and Ethical Responsibilities of Computing programme is in its seventh year as of August 2026.
The committee itself draws on that lineage: Breazeal and Klopfer, RAISE’s director and one of its co-directors, both sat on it, with Klopfer as co-chair. The report was written while MIT’s Task Force on the Undergraduate Academic Program was separately considering broad curricular change, and the committee points to that work as the vehicle for several of its recommendations. Its own proposal is that the questions do not close: it asks MIT to fund a standing committee and an implementation team rather than settle on a single policy, on the grounds that, as it puts it, “no one seems to have it all figured out”.
Sources: MIT, Report of MIT’s Ad Hoc Committee on AI Use in Teaching, Learning, and Research Training; MIT AI and Education; MIT AI and Education, Appendices; MIT, letter from President Sally Kornbluth; MIT, committee charge letter; MIT Faculty Governance; The Tech; MIT Quality of Life Survey; MIT News; The Washington Post
Featured image: photo by Gera Cejas on Pexels (free Pexels license).
Become a Sponsor
Our website is the heart of the mission of WINSS – it’s where we share updates, publish research, highlight community impact, and connect with supporters around the world. To keep this essential platform running, updated, and accessible, we rely on the generosity of you, who believe in our work.
We offer the option to sponsor monthly, or just once choosing the amount of your choice. If you run a company, please contact us via info@winssolutions.org.
I specialize in sustainability education, curriculum co-creation, and early-stage project strategy. At WINSS, I craft articles on sustainability, transformative AI, and related topics. When I’m not writing, you’ll find me chasing the perfect sushi roll, exploring cities around the globe, or unwinding with my dog Puffy — the world’s most loyal sidekick.
