By now, most families have had some version of “the AI conversation.” As we noted in our recent post on creative thinking and standardized testing, the share of high school students using generative AI in their schoolwork has climbed past 80%. So the question has shifted from “Will my student use it?” to “What are they giving up when they do?”
That’s the question Academic Approach founder Matthew Pietrafetta and Andrew Ferguson took on in the latest episode of our podcast, Beyond the Test. Matthew has a name for what worries him most, borrowed from a recent MIT report: cognitive surrender.
What is cognitive surrender?
Cognitive surrender is the term MIT’s Ad Hoc Committee on AI Use in Teaching, Learning, and Research Training used in its August 2026 report on AI and education for what happens when a student falls back on AI at the first hint of struggle and stops working through the problem. (You may also hear it called cognitive offboarding or, as Matthew puts it, cognitive outsourcing.) The worry isn’t that chatbots are sometimes wrong. The committee warned that a chatbot’s right answer can create “the illusion of learning” while the student skips the thinking: the part where they analyze, synthesize, evaluate, and make a judgment call.
That thinking is exercise for the brain, and research from the MIT Media Lab gives a glimpse of what the difference can look like. In a 2025 study called Your Brain on ChatGPT, 54 students wrote essays while researchers recorded their brain activity with EEG. The students who used ChatGPT showed the weakest connectivity between brain regions, and those who wrote with no tools showed the strongest. (The study is a preprint with a small sample, so it’s early evidence rather than the final word, but it points in the same direction as the MIT committee’s concerns.)
It’s not just about the essay
If cognitive surrender cost students one assignment, it wouldn’t be much of a concern. The bigger loss is everything that comes bundled with the struggle. Working through something hard is how students practice planning, tolerating frustration, putting off gratification, managing their time and emotions, and knowing when to ask a teacher or tutor for help. Educators call this productive struggle, and it matters most during adolescence, when students are developing all of these skills at once.
Andrew borrowed a line from Mike Tyson to describe what he’s seeing: everybody has a plan until they get punched in the mouth. Students used to try to avoid the punch (the bad grade, the hard problem). Now it seems like they’re trying to avoid making the plan at all. And without a plan, there’s nothing to learn from when things go sideways: no “I should have started earlier,” no adjustment for next time.
There’s an ownership problem, too. Matthew passes along something his own mentors told him: nothing worth having comes easily. If ChatGPT hands over the essay, there’s no “sweat equity” in it. The student can’t really claim it, and they’re less likely to value it. (In the MIT Media Lab study, students who used ChatGPT reported the lowest sense of ownership over their essays.) The real question, as Matthew frames it, is who your student wants to become.
So should students never use AI?
Not at all, and Matthew is the first to say so. The morning they recorded the episode, he used Waze to find a friend’s house instead of unfolding a map. He knew he was outsourcing a mental task, and he decided the convenience was worth it. (Learning the street layout of his friend’s town wasn’t going to make him a better person.) He wouldn’t make the same call on writing a paper or working through a difficult novel, because those are the tasks that build the skills he cares about.
The trouble is that most students aren’t making a choice at all. AI is immediate and gratifying, so they go there by default, and no one has talked with them about what they’re giving up. In Matthew’s words, they aren’t yet “informed consumers.”
Even expert tutors use AI, but with guardrails. Matthew recently used ChatGPT to pull up the steps for resetting a graph on a TI-Nspire calculator (faster than a Google search, and without the ads), and Andrew uses it to generate extra practice questions. What makes that work is that both of them know the material and the student well enough to judge what comes back. Students are still building that evaluation skill, and they build it by doing the very work AI would do for them.
What can families do?
AI isn’t going back in the bottle. (“The genie’s already out of the bottle,” as Matthew puts it, and regulation is unlikely to arrive in time to help this generation of students.) The most useful thing families can do is help students make the choice on purpose. Here are a few questions to ask together before they open a chatbot:
- Is this a task that builds a skill I need? Reading, writing, planning, and working through a hard problem are workouts. Looking up an address isn’t.
- Could I tell if the answer were wrong? If not, the tool isn’t saving time. It’s hiding the problem.
- Whose work is this? If your student couldn’t explain it or redo it without help, it isn’t really theirs yet.
- What happens on test day? The ACT and SAT are taken alone, with no chatbot in the room. Whatever thinking a student hasn’t practiced is thinking they won’t have when it counts.
How to Practice: The next time your student gets stuck, resist the urge to hand over the answer (or to let a chatbot do it). Ask what they’ve tried, what they might try next, and how they’ll know it worked. These are the same questions our tutors ask, and over time students start asking them of themselves.
Where we stand
At Academic Approach, the goal of tutoring has always been independence, not dependence. Matthew describes his job as trying to “write myself out of a job”: through modeling, scaffolding, and a steady back-and-forth of questions, students internalize the problem-solving dialogue until they can run it on their own. That’s the opposite of what happens when a chatbot simply hands over the answer, and it’s the work we plan to keep at the center of our instruction.
Want to hear the whole conversation? Listen to Beyond the Test, Episode 3: Cognitive Surrender, or find the show on Apple Podcasts and Spotify. And if you’d like help building the thinking skills that last well beyond any one test, contact us today.