MIT's Report on AI in Teaching and Learning

The Massachusetts Institute of Technology (better known as MIT) released a report last month sharing the findings and recommendations of their Ad Hoc Committee on AI Use in Teaching, Learning, and Research. I've read a lot of university reports on AI over the last few years, and most of them are forgettable. This one, however, is fantastic. It's very MIT, in that it's grounded in the MIT context, which is not a typical college or university context, but it takes such an informed, realistic, and practical take on AI and higher ed's response to AI that I think anyone can benefit from reading the report.

As my Norton Guide to AI-Aware Teaching co-author Marc Watkins pointed out, the MIT report uses the term "AI-aware" repeatedly, and I found out via LinkedIn that at least one person on the ad hoc committee participated in this summer's community read of The Norton Guide hosted by Perusall. So I'd like to think that our book had some impact on the recommendations shared in the report! And those recommendations are solid!

1. Adapt educational processes for an AI-aware world, starting with revisiting course goals and objectives, then leaning into alternative assessments (such as oral exams and semester portfolios) and experience learning and in-person social learning
2. Center people, community, and the residential experience through new shared norms and experiences along with responsible and ethical AI use, including appropriate disclosure of AI use
3. Build processes, teams, and tools for continuous reflection, iteration and improvement, such as providing faculty with the expertise and time they need for AI-aware course and assessment redesign

If you're interested in my views on some of the details in the report, I encourage you to read and respond to my public annotations of the report via Hypothesis. For my general thoughts about the report, here are a few...

As my University of Mississippi colleagues Josh Eyler and Emily Donahoe (and others) have noted, there's a welcome emphasis in the MIT report on alternative grading practices, especially competency- and mastery-based assessment and portfolio grading. I've talked to many individual instructors who are experimenting with non-traditional approaches to grading as a response to the assessment challenges posed by AI, and I think it's a positive development that an institution like MIT would argue for greater use of such approaches. (In her blog post this week, Emily compared and contrasted MIT's recommendation to grading moves made recently by other institutions, namely Harvard and the University of Michigan.)

While I appreciate that the report reflects the context and values of the MIT education context, that's a context that clearly doesn't involve online education in a significant way. There's a lot of emphasis on the value of residential education in the report, as you might expect, but that emphasis might be off-putting to those teaching at institutions where onsite education is just one of multiple modalities. Some of the recommendations in the report that make a lot of sense for a residential institution won't translate well to other institutions. I think that's fine, as long as the reader knows that going in.

The report mentions a few rather innovative uses of AI to support learning, the kind of "treadmill in a gym" uses that I like to explore and share with faculty. Specifically, there's mention of using customized AI chatbots to create learning experiences for students and the use of AI vibecoding to design interactive online learning materials for students. These aren't common practices among the faculty that I know, so is MIT particularly ahead of the curve on the use of AI in teaching? Maybe. I mean, it's MIT. On the other hand, the report also notes that "students report that the AI guidance they receive from instructors is often confusing and unclear..." and often varying "widely" across courses. This is a very common student report across higher ed, so perhaps MIT has a very similar mix of faculty as other institutions when it comes to ways they're responding to AI.

Finally, many of the recommendations in the MIT report are right in line with our advice and examples in The Norton Guide to AI-Aware Teaching. So, at the risk of being a shill for my own book, if you're interested in practical advice for implementing MIT's AI recommendations, maybe check out the new book?

Around the Web

"There's so much to read!" That's what I said out loud to myself earlier this week as I kept finding interesting reports and posts and articles relevant to my current projects. That says to me that it's time for an "Around the Web" edition of the newsletter!

• "Handwritten Notes Aren't a Magical Solution" - This post by Sara Misgen on the Engaged Learning Collective blog is absolutely fantastic. I've been fascinated by notetaking processes for years. See, for instance, my attempt at a lit review and argument for using sketchnotes back in 2014. Sara notes correctly that there's been a renewed interest in handwritten notes as a response to AI, often as part of a tech-free classroom design of some kind. Sometimes the calls for handwritten notes refer to that 2014 study "The Pen Is Mightier Than the Keyboard" study by Mueller and Oppenheimer. Sara starts with that study and situates it in the larger body of research on notetaking that has come after it. She makes a very strong case for "it's complicated" as a way to sum up the research on notetaking! Her post is a must-read for anyone interested in supporting student notetaking. Also (spoilers), you'll get to hear Sara on my podcast sometime this fall.
  ​
• ​The PAIRR Project's Contradictory Feedback Prompts - Last year around this time, I had Marit MacArthur and Anna Mills on the podcast to talk about the PAIRR Project. PAIRR stands for Peer & AI Review + Reflection, and it's a way of thoughtfully integrating AI feedback in the classic peer review process often used in writing courses. I've shared the PAIRR Project in just about every workshop I've led since then, and the PAIRR team continues to explore this approach to teaching through scholarship of teaching across multiple institutions. Their upcoming experiment involves prompting the AI feedback tool to provide contradictory feedback to students, that is, "mutually incompatible suggestions" for writing revision. They don't have results to share, but I love the idea. It's a great example of a way to help students adopt better mental models for working with AI, as I wrote about in this space (and on my blog) recently, specifically using AI as a kind of possibility machine.
  ​
• ​Departmental AI Discussion Toolkit - Jenae Cohn is the executive director of the Center for Teaching and Learning at UC Berkeley, and like a growing number of CTL folks, she's working more and more at the departmental level to help faculty respond to AI. Y'all know I think higher ed has a coordination problem when it comes to AI. While it's important that individual faculty make informed, intentional, and AI-aware choices about course design, our students don't take courses in isolation. It's critical that departments and programs work toward some kind of coherence in their response to AI. Jenae knows this, too, and based on her work with departments, she has drafted this discussion toolkit for helping departments work toward that coherence. She's just getting started developing the toolkit (and I'm hoping to contribute some of what I'm developing at UVA), but there's already a lot there to inform good meso-level discussions about AI and teaching.

The Norton Guide to AI-Aware Teaching

My new book, The Norton Guide to AI-Aware Teaching, co-authored with Annette Vee and Marc Watkins, is now available as an ebook! It's available through major online retailers like Amazon and Barnes & Noble. It's also free for faculty who have adopted a W.W. Norton textbook, so if that's you, ask your Norton rep. If you want to wait for the paperback, it will be available in late September 2026, and you can preorder it now.