AI Research

What MIT's AI Report Means for a Secondary English Class

In August 2026, a committee at MIT published a long report on AI in teaching and learning. It's written for a university, and much of it's about problem sets, research labs, and residence life. But its central conclusions translate almost directly to a Grade 10 English class, and they arrive at the same place this site does.

Who Wrote It, and Why It's Worth Your Time

The committee included undergraduate and graduate students, faculty from every school at MIT, and staff from units such as the Libraries and the Teaching and Learning Lab. It spent five months meeting, researching, and holding listening sessions, and it reports that its members came in with no fixed thesis and left with a strong shared view.

That matters because MIT isn't an institution nervous about technology. When a university this closely tied to the birth of AI concludes that assessment must change and that people have to stay at the centre of learning, it's worth listening. It's also worth being clear about the limits: this is university guidance, and in a secondary school your board or district sets the outer policy. What follows is a translation, not an endorsement of anything on this site.

1. The Real Risk Is Lost Learning, Not Cheating

The report's sharpest line isn't about academic integrity at all. What should worry us most, the committee writes, is that "many uses of AI deprive students of the opportunity to learn." It describes learning as productive struggle and warns that getting the right answer from a chatbot can create an illusion of learning.

For English, that reframes the whole conversation. The question isn't only whether a student cheated. It's whether the assignment still makes the student do the thinking it was designed to build.

2. Stop Policing, Start Designing

The committee recommends against relying on AI detectors. It warns that they can mistake the writing of non-native English speakers and neurodivergent students for AI, that policing builds an adversarial atmosphere, and that MIT's own disciplinary committee doesn't treat detector output alone as sufficient evidence. In its words, mutual suspicion "is no foundation for a healthy classroom."

What it suggests instead will be familiar: early drafts written by hand in class, regular checkpoints that show a project developing, feedback at several stages rather than only on the final product, and version histories submitted with the work. That's process evidence, the same case this site makes in the piece on detectors.

3. Don't Just Move Everything into Timed Writing

This is the report's most useful warning for secondary teachers. Many instructors are responding to AI by shifting more marks to in-class exams. The committee cautions that timed, high-stakes evaluation cuts the deliberation students can put in and weakens their reason to invest in longer, harder work.

Instead it recommends oral exams, portfolios, and out-of-class assignments paired with in-class conversations. In an English class, that's a balance, not a ban: some thinking gathered over time, some done in the room, and a final product that must match both. It's the reasoning behind the 40-40-20 model.

4. Tell Students Why, Assignment by Assignment

Students told the committee that AI guidance was confusing and varied from one instructor to the next. The report's answer is clarity with a reason attached: students should know when, where, and why AI is prohibited, allowed, or required, and the reason should be tied to what the task is meant to teach.

Your board or district decides the rules around AI. Within those rules, the explanation for each assignment is yours to give, and it's where students learn the most. Naming the level of AI use a task allows, and why, is exactly what the Audit helps you do.

5. Augmentation, Not Automation

The committee's principle is that AI should extend what students can think about and do, never replace the thinking itself. It names the danger "cognitive surrender", which is reaching for AI at the first sign of struggle. It wants students to learn when a model is likely to invent things, when not to use AI at all, and how to disclose honestly what AI contributed.

In English, that's teachable. A student who can explain what a chatbot did, what they did themselves, and how they checked the result is learning to use AI as a tool rather than a ghostwriter.

6. Keep the Room Human

The report worries that AI is quietly replacing the human parts of learning: study groups, office hours, arguing an idea out with a peer. It recommends that every course include structured, in-person social learning, such as discussion where individual participation is assessed and feedback conversations built around a shared rubric.

English has always done this well. Literature circles, conferences, oral defences of an argument, and seminars are some of the most AI-resistant things we teach, because they can only happen in the room.

Where It Doesn't Translate

Plenty of the report is specific to a research university: undergraduate research programs, residence life, grading reform, and campus AI platforms. Its point about access does carry over, though. Some students can pay for far more powerful AI tools than others, so a task that rewards paid access isn't a fair task. That's worth keeping in mind whenever an assignment invites AI in.

Where to Start

With one assignment you already use. The Audit asks eight questions about it and shows how much of the thinking a chatbot could do for your students, and the first thing worth changing.

Questions Teachers Ask

Where This Comes From

Everything above is drawn from one publicly available report. Read it in full rather than relying on any summary, including this one.

• 1
  MIT Ad Hoc Committee on AI Use in Teaching, Learning, and Research TrainingReport, August 13, 2026. Available at aiandeducation.mit.edu.

Developed by George O'Toole: over 35 years in secondary English, Head of Department, Educational Technology Coach, Faculty of Education instructor. Meet George