As someone who teaches Python programming for a living, I’ve spent the last few years wrestling with the educational implications of AI. I’m changing everything I do to adjust to our new AI reality, experimenting with new ideas, including my AI-based Socratic tutor (https://practice.lernerpython.com/). I keep what works, throw away what doesn’t, and then try the next thing. I’ve never been more challenged as a business owner. And yet, I’ve never been more excited about being an instructor.
What about universities? I’ve said for a while that they’ll also need to change, but that they cannot do so nearly as quickly as I can. They’re big and bureaucratic, and have to answer to donors, staff members, students, parents, and governments. It’ll take years for them to figure out what they want to do, and how to do it.
And then, last night, I received e-mail from Sally Kornbluth, the president of MIT, with a message addressed to all MIT students, faculty, staff, and alumni. The subject, “AI and education: A watershed moment for MIT,” was quite the understatement.
Kornbluth announced a report from an ad-hoc committee on AI in education (https://aiandeducation.mit.edu/report/). It’s the clearest, deepest, and most profound take I’ve yet read on the subject. It doesn’t hold back, spelling out the good and the bad. I’m still absorbing the full impact of what they wrote — and they wrote a lot. But the report — which I expect will be given formal backing in the near future — has, more or less, called for a complete revolution in how instruction works. And they’re willing to take the lead in trying new approaches, technologies, and assessments, both for their own future and for the future of university education.
Maybe a university can’t change as quickly as my one-person business. But the report sounds urgent and focused, and I think we should expect MIT to change as quickly as any large organization can. The report even recognizes the slow pace at which curricular changes are normally made, and encourages departments to allow for curricular experimentation without onerous bureaucracy.
Among the many smart, important points they make:
- Classes are going to change. Instructors will need to re-assess what people need to get out of a course, and think about how AI can be integrated into it. Whether that means developing new course-specific tools or having students use AI to advance their knowledge and understanding will depend on the course and the instructor. But they are actively encouraging instructors to think deeply about what they are really trying to teach, and what tools might help them to achieve that more easily.
- Embracing apprenticeship. You could argue that graduate school (and to a lesser degree, undergraduate studies) are already an apprenticeship program, in which aspiring researchers learn from their more experienced professors and peers. AI threatens this model to some degree, enticing researchers to use agents to accomplish a task, rather than bringing on someone who might take longer or make mistakes. As they say in their report, “Using research as an opportunity for learning-by-doing may produce seeming ‘inefficiencies,’ but that’s a feature, not a bug.” Struggling and working through problems is an inherent part of the learning process.
- Education is about the process of learning. I keep meeting people (students and professionals) who say that they ask AI to write papers and essays for them, because they “know it anyway,” and this is just a faster way of producing the same text they would have written on their own. This is, of course, nonsense — as the saying goes, “writing is thinking,” and anyone who writes knows that as you struggle over how best to make your argument, you’re learning more than any classroom could teach. As the report says: “Getting the right answer from a chatbot can create the illusion of learning – but it can also trigger ‘cognitive surrender’, where students fall back on AI at the first hint of struggle.” They similarly say, “All of us who teach at MIT will need to be prepared to help students understand both that the process of education is necessarily a productive struggle, and that the most important product of their education is not a GPA or a diploma but themselves: their personal growth and intellectual maturity and the development of their own imagination, insight, and judgment.”
- Assessment will need to change. The report says, “MIT should take this opportunity to consider what role grades play in our overall system, and if the current approach could be improved.” The old standards of homework assignments and exams might not be the best ways to know whether a student has really mastered the material. I’ve been saying for a while that it might be time to bring back oral exams, either alone or alongside project-based learning. The report discourages the use of AI detectors, which have a history of being unreliable and discriminatory. Plus, “stepping up ‘policing’ around AI use builds an adversarial atmosphere of distrust between instructors and students, which understandably hurts students’ motivation and morale.”
- Constructionism has its moment? I feel like this is a moment in which all of the educators who spoke of project-based, constructivist, and constructionist approaches for decades might finally have their moment in the sun. The report points to the potential for individualized attention and instruction that AI can uniquely provide — and which would go hand-in-hand with the project-based approach. Seymour Papert developed constructionism at MIT, calling for a revolution in education around personalized, project-based, technological learning. I’m an MIT alumnus, a constructionist at heart, and I did my PhD in learning sciences under one of Papert’s students. So watching Papert’s own institution consider a revamp of education around his principles is, for me, delicious.
- Ethics are crucial. We’re now in an era where anyone can generate a decent-sounding article within minutes. The report emphasizes that a culture of transparency around the use of AI is and will continue to play a critical role. This is true not just for students, but also for instructors and researchers, who will be expected to clearly state where and how they used AI in their work. The report says, and I believe rightly so, that using AI is just fine — but you need to be specific about how you use it, and what you did with it. They also point to the need for equitable access to AI, ensuring that all members of the MIT community can use this technology, not just those who can most easily afford it.
The report also emphasizes, repeatedly, the importance of human interactions in all of this. AI, more even than phones and the Internet, has a tendency to push people to hole up by themselves, rather than interact with others. (The report points to fewer in-person study groups, and a drop in the number of students coming to office hours.) And it is those interactions that are not only at the heart of research, science, and learning, but of human existence. They stress the need for incorporating “social learning” into projects, and to encourage people to work together. As they say, “Instructors should intentionally structure such interactions to achieve desired learning objectives and maintain quality, even in large classes.”
This report isn’t the last word on AI and education, and it isn’t meant to be. A year from now, we’ll probably see parts that were prescient, and other parts that were misguided (if well meaning). But it’s the first time I’ve seen a major university announce a plan to reshape the entire university’s educational approach around our new reality. This is an ad-hoc committee, and thus doesn’t have any official standing. But the fact that MIT’s president sent it out to every member of the MIT community tells me that this is way beyond a simple committee. I expect that MIT will put serious effort (and money) behind the changes needed to see these things through.
If you’re an educator, then you owe it to yourself, and to your students, to read this report. And then to re-assess what you do, starting with the principles that MIT is using to evaluate itself, as the first stage of what’ll likely be a revolutionary transformation.