MIT Study Shows Assigning Essays Triggers Student "Brain Rot"

The big question isn’t how to put AI in schools but how to redesign instruction to support the reality of AI

Instead of endlessly policing plagiarism, punishing students, or lamenting that our own assignments are rotting students’ minds in the age of AI, we need to rethink the task itself. Like wise builders who reinforce or relocate homes when the climate shifts, educators must design assessments that are resilient, relevant, and responsive to the new intellectual environment.

---

A new study out of MIT about the impact of ChatGPT use on one’s brain has triggered some startling headlines.

It seems many of us, maybe even nearly a billion of us, are doomed.

But what a minute, let’s look more carefully at the study (you can also read it and use your own critical thinking rather than relying on clickbait headlines that are rapidly shared by others on social media by others who haven’t read it).

The study was conducted with 54 participants aged 18–39 over three sessions (plus a fourth crossover), It compared three groups: those writing essays unaided, those using Google (sans AI overviews), and those using ChatGPT. Researchers measured brain activity via EEG during 20-minute SAT-style essay tasks, analyzed essays with NLP techniques and human grading, and interviewed participants.

Results show that the brain-only group displayed the strongest, most distributed neural connectivity—especially across alpha, theta, and delta bands linked to memory, creativity, and attention—while Google users showed moderate engagement, and ChatGPT users exhibited significantly weaker connectivity. ChatGPT users also produced formulaic, homogenous essays, reported lower ownership, and struggled to recall their own writing even minutes later. Over time, their neural engagement and originality declined, a phenomenon the authors term “cognitive debt.”

It’s easy to criticize the study’s methodology. It has a very small sample size (and all of the participants are from top universities), it only test a limited number of them 4 months later, and weaker brain activity could signal a reflective state).

To me, however, the larger problem is not the design of the study, but that people are drawing the wrong conclusion: that ChatGPT use in general causes cognitive declline.

I see two more signficant limitations than aything associated with its design.

(1) The study was limited to paper writing — if you have something else such as ChatGPT write your paper, you’ll learn less.

So, unlike many editorials about this study that do generalize to a general loss of thinking, this study does not.

(2) It really states the obvious — if you have someone do your work for you, you won’t think while doing the work and you won’t learn anything from the work being done (the study notes that when an LLM writes a paper the people who used the LLM to write the paper do not remember what is in the paper). It’s very similar to the study in Turkey that found that when students turned their coding projects over to an LLM that they didn’t learn how to code.

To me, this is a “no shit” moment. If something or someone (your parent, friend, tutor), does your work then you won’t learn much of anything from it.

So what do we do?

It seems there are two options.

(1) We can rail against the reality that generative AI exists, keep assigning essays for students who will use ChatGPT to do them and rot their brains, and spend millions of hours trying to catch kids “cheating.” We can also get on social media and talk about how the assignments we give cause brain rot.

Here are some alternative headlines —

This is like building a home on a coastline known for destructive hurricanes—then acting shocked and outraged each year when your house gets flattened. If we keep assigning traditional essays knowing students can and will use ChatGPT to generate them, we’re choosing to build in a danger zone and then blaming the storms when the structure collapses.

(2) We can restructure assignments and stop giving assignments that trigger brain rot in the world of AI. For example, here are some ways to reduce reliance on essay writing and build alternative assesssment into classroom instruction so that when students use ChatGPT to prepare they are likely to increase their critical thinking.

Debate

• Integrate “AI vs Student” sparring rounds
  Ask students to write their own argument, then have ChatGPT generate a counterargument. The student must rebut it live, practicing critical engagement with AI-generated reasoning.

• Hold “Tagline Spotlight” speaking drills
  Students generate concise claim statements (taglines) and practice supporting and defending them in under 30 seconds, improving their clarity and logical reasoning with or without AI help.

• Replace essays with oral defense panels
  Have students research a topic (with or without AI tools) and then respond to a panel of student or teacher questions, forcing them to defend their ideas in real time.

• Run mini crossfire sessions
  Teach students to engage in 1-on-1 rapid questioning sessions based on prepared arguments, helping them probe for contradictions and gaps—skills that ChatGPT can help simulate but not replace.

• Socratic Seminars with student-led questioning
  Students generate their own discussion questions (possibly with ChatGPT’s help), but must lead and sustain a live discussion using evidence, listening skills, and follow-up reasoning.

• One-minute impromptu speeches
  Students are given a prompt with limited prep time (e.g., 3 minutes), where even if they’ve used ChatGPT previously, they must synthesize, organize, and present ideas on the spot.

• Evidence showdown
  Students each bring in one piece of evidence (possibly AI-sourced), but must explain why it’s credible, relevant, and strategically useful—building source literacy and application skill.

• “Make it your own” paraphrasing drills
  Give students a ChatGPT paragraph on a topic, and ask them to rewrite it in their voice, with clearer logic or better examples. Score them on originality, clarity, and improvement.

• Framing battles
  Have students take the same set of facts or evidence and frame them in two different ways (e.g., pro vs. con, historical vs. modern, individual vs. societal impact)—a skill even the best AIs can’t do without user input.

• Simulation or role-play debates
  Put students in the shoes of historical figures, policymakers, or stakeholders. Even if AI helps generate background info, the student must embody the character’s reasoning and make strategic choices in dialogue.

  
  
  
  Project Based Learning

• Design and Pitch a Solution to a Real-World Problem
  Students use ChatGPT to research background and brainstorm solutions, but must design a prototype, write a proposal, and pitch their idea to a panel—showing understanding, creativity, and real-world application.

• Create an Explainer Video or Podcast
  Even if AI helps script or summarize, students must present the material in their own words, demonstrating clarity, voice, and conceptual mastery in an engaging, original format.

• Build a Public Resource or Toolkit
  Students create a guide, toolkit, or infographic for a real audience (e.g., teens learning about climate change or voters understanding a ballot measure). ChatGPT may help with background, but students must curate, adapt, and personalize the content meaningfully.

• Community-Based Research or Action Projects
  Students identify a local need, conduct interviews or surveys, and propose or enact a solution. While AI can assist in research or drafting communications, the student must collect, synthesize, and apply live data from their community.

• Design a Role-Playing Scenario or Simulation
  Students build a mock international summit, historical simulation, or ethical dilemma game. They can use ChatGPT to generate initial materials, but must adapt content, create roles, facilitate interaction, and reflect on outcomes.

• Curate an Exhibit or Interactive Timeline
  Students develop a digital or physical display on a theme (e.g., revolutions, pandemics, tech ethics). Even if AI aids research, the student must select, organize, and narrate the content to tell a coherent, compelling story.

• Launch a Passion-Based Campaign or Business
  Students use AI to help brainstorm names, research markets, or write mock press releases—but the project must include real-world testing, design, budgeting, or outreach components driven by the student’s goals.

• Debate from Within the Project
  After completing a group project (e.g., a proposed city redesign or economic policy), students defend their plan in a structured debate—explaining trade-offs, anticipating objections, and demonstrating decision-making.

• Create a “Humans + AI” Learning Artifact
  Students must show both what the AI contributed and what they did to improve, refine, or challenge it. For example, a science explainer where they mark AI-written text and add their own annotations, improvements, or graphics.
  

We can even learn from the students. Here are some ways that students use ChatGPT that are likely to increasing their critical thinking (at least most of them).

While I don’t have a 200 page study to back it up, it seems quite obvious to me that all of these activities, even when using ChatGPT, will increase critical thinking. In fact, the usage of ChatGPT (or any other similar LLM), may signficantly contribute to the increase.

This is why the big question isn’t how to put AI in schools but how to redesign instruction to support the reality of AI.

Instead of endlessly chasing plagiarism, punishing students, and now saying our graduates’ brains rot under our tutelage as a result of the assignments we give, we need to rethink the assignment itself—designing assessments that are resilient and relevant to the new environment, just as smart builders reinforce homes or relocate them entirely when the weather changes.

Yes—students still need to learn how to write. Writing helps us clarify our thinking, develop persuasive reasoning, and communicate with depth and precision. But writing is also evolving. Increasingly, it will involve writing with AI—using tools to collaboratively, brainstorm, refine, and communicate more effectively. In writing courses especially, we should still emphasize revision, feedback, and authentic expression. But that doesn't mean we should cling to the traditional essay as the dominant form of assessment.

In fact, when students are assigned essays merely to fulfill a requirement—and when they know a chatbot can produce a passable five-paragraph response—we risk encouraging a shallow approach to learning. If students aren’t thinking deeply, and we keep assigning essays as if they are, we become complicit in their “cognitive offloading”—a disconnect between effort and outcome that hollows out the purpose of writing. In such cases, the exercise doesn’t strengthen their minds; it bypasses them.

In both debate and project-based learning (PBL), writing plays a critical and multifaceted role, but it looks quite different from the traditional essay. Instead of serving primarily as a summative assessment, writing in these contexts functions as an interative tool for thinking, organizing, persuading, and communicating. In debate, students engage in purposeful, high-stakes writing that helps them build arguments, anticipate counterarguments, and respond strategically under time pressure. One of the primary places where debate students write is in constructing their speeches. This includes pre-written constructive speeches, where they develop claims, organize evidence, and carefully craft impacts. Febuttals and summaries, which are more spontaneous, require written preparation in the form of “blocks” or outlines of likely responses. These processes build important writing skills like conciseness, clarity, and logical structuring.

Beyond speeches, debate students spend significant time writing and organizing their research. They create full cases, evidence files, and annotated briefs—skills that develop their ability to summarize, paraphrase, and cite sources responsibly. During actual rounds, students also write in real time through a process called “flowing,” where they take notes to track arguments, responses, and dropped points. This note-taking is not passive—it trains students to write quickly, distill complex ideas, and map the logic of the round. Additionally, debate writing often includes the creation of short, strategic texts like judge summaries or "frontlines," which are quick pre-written responses to anticipated arguments. These tasks hone students’ awareness of audience and rhetorical framing, as they must adapt their writing to the judge’s preferences and the evolving dynamics of the round.

Project-based learning immerses students in longer-term writing experiences tied to real-world problems and authentic audiences. At the beginning of a PBL unit, students typically engage in writing by planning their projects. They develop driving questions, write proposals, and outline research plans. These foundational writing tasks help students set goals, clarify their purposes, and take ownership of the inquiry process. As the project progresses, students write to learn and document. They take research notes, summarize findings, log data, and create annotated bibliographies. These activities strengthen information literacy and academic writing habits while keeping the focus on real-world application.

Throughout the project, students also engage in reflective writing. This might include process journals, learning logs, or group collaboration notes. Such writing builds metacognitive skills and supports deeper learning by prompting students to think about how they are learning—not just what they are learning. At the culmination of a project, students often write to communicate their ideas to external audiences. This could take the form of a report, blog post, multimedia script, presentation pitch, brochure, or formal letter. These products emphasize genre awareness, real-world purpose, and the need for clarity and persuasion. Finally, feedback and revision are integral to the PBL writing process. Students review one another’s work, respond to teacher comments, and make improvements. This iterative cycle builds resilience, editorial judgment, and collaborative writing skills.

Although debate and PBL differ in structure, they share common ground in how they treat writing: not as a static product but as an active process of reasoning and communication. Debate sharpens quick-thinking persuasive writing under pressure, while PBL encourages sustained, reflective, and audience-centered writing over time. In both, writing is a means of navigating complexity, building arguments, clarifying ideas, and connecting with others. Importantly, these approaches help students develop writing skills that are relevant for the world beyond school—skills like making a case, responding to challenges, adapting to different genres and audiences, and revising in response to feedback. Rather than asking students to write for the teacher alone, debate and PBL ask them to write for impact.

Both debate and project-based learning treat writing not as a final product to be graded and forgotten, but as a dynamic tool for thinking, exploring, and making meaning. In these settings, students write to clarify their own ideas, to communicate effectively with others, and to persuade real audiences—not just to complete a checklist or fulfill an assignment. This reframing of writing transforms it into an authentic, engaging, and interdisciplinary act. Writing becomes authentic because it is tied to real outcomes: winning a round, informing a community, or solving a meaningful problem. It becomes motivating because students know their words matter and must be timely, impactful, and well-constructed. And it becomes interdisciplinary because it draws on research, critical thinking, collaboration, and creativity—pulling together skills from across the curriculum. In both debate and PBL, writing comes alive as a purposeful, high-stakes process, helping students develop into agile thinkers and communicators who can adapt to complex tasks and audiences.

This kind of writing is more difficult to turn over to an AI because it demands deep personal engagement, real-time adaptation, and iterative thought—all qualities that resist automation. When students write to persuade a judge, respond to peer critique, or propose a novel solution to a real-world issue, they must wrestle with ambiguity, revise based on feedback, and tailor their message to specific, unpredictable audiences. These are not static prompts with predictable formats but evolving challenges that require voice, judgment, and situational awareness. AI may assist at the margins—offering phrasing suggestions or helping gather evidence—but it cannot substitute for the student's own thinking in the heat of preparation, dialogue, and reflection. Because the writing is tied to live stakes and evolving contexts, outsourcing it would not just be inauthentic—it would be ineffective.

The world is changing fast. Generative AI is already transforming how people work, with many adults facing career shifts, job loss, or the challenge of learning to work alongside AI. Compared to that, educators have a simpler—but still urgent—task: rethinking how we assess learning. We don’t need to eliminate writing, but we do need to move beyond default essays and “reasearch papers” and toward experiences that require students to think: oral defenses, live debates, collaborative problem-solving, real-world design, and ethical reflection. Our goal isn’t to preserve the way we were taught—it’s to prepare students for the world they’re entering.

We need to admit that we cannot control whether GAI exists and if students use it, and instead structure teaching and learning in a way that adapts to the reality of the new world.

These new approaches don’t just discourage cognitive offloading—they actively build the skills students will need to thrive. In the AI era, value will be determined by the ability to use intelligent tools to solve problems at scale. That means prompting clearly, evaluating outputs critically, and refining ideas with human insight. A student who uses AI to brainstorm a startup idea, but then tests it with real users and adapts it based on feedback, is learning how to bridge AI capabilities and human judgment. A student who uses AI to explore both sides of a controversial issue, and then defends a position in live debate, is developing the fluency to lead in high-stakes, real-world conversations. These are not just school skills—they are future-proof capabilities in medicine, policy, engineering, disaster response, and beyond.

This is why the question isn’t how to add AI into schools—it’s how to redesign instruction for a world where AI is already here. Simply inserting AI into old lesson plans or ranting about “cognitive offloading” that results from the assignments we choose to give misses the point. What’s needed is a shift in mindset: from teaching content and assigning essays to building capability, from assigning tasks to cultivating thought. In a world where machines can generate answers instantly, our job is not to compete—it’s to help students learn how to guide, challenge, and collaborate with those machines. That means letting go of static essays as the default, and embracing instructional designs that develop agency, originality, and wisdom. If we rise to that challenge, we’ll do more than engage students—we’ll equip them to shape the future.

We need to stop contributing to “brain rot” and change our assigments.