The video argues that while headlines claim AI is making college students unable to read, the reality is more nuanced. Research—including an MIT EEG pre‑showing reduced brain connectivity and cognitive “amnesia” when using AI, a Carnegie‑Mellon/Microsoft study linking high AI confidence to lower critical thinking, and larger surveys finding a negative correlation between frequent AI use and critical‑thinking scores—shows that indiscriminate reliance on AI offloads thinking and can cause lasting cognitive deficits, especially in developing minds. However, studies also reveal that when AI is used critically (questioning, editing, revising) it can enhance learning and reduce mental fatigue. The problem is not the tool itself but how it is deployed: passive consumption erodes cognition, while active engagement builds it. The speaker urges a shift from general‑purpose LLMs toward specialized, purpose‑built AI (e.g., Socratic‑style educational assistants) that forces users to think harder, calls for more rigorous, longitudinal research across ages and learning styles, and stresses that addressing AI’s cognitive impact is an urgent, interdisciplinary design challenge—not merely a PR or banning issue.
1. MIT preprint titled “Your Brain on Chad GBT” by Cosmina et al. (2025) studied 54 participants aged 18‑39 using EEG headsets while writing SAT‑style essays.
2. Participants were assigned to three groups: one used Chad GPT, one used Google search, one used no external tool.
3. Chad GPT users showed the weakest neural connectivity and lowest brain engagement in regions linked to memory, attention, critical thinking, and creative reasoning.
4. Google‑search users showed moderate neural engagement; participants who wrote without any aid showed the strongest, most distributed neural activity.
5. 83% of Chad GPT users could not quote a single line from the essays they had just written, indicating cognitive amnesia.
6. In a follow‑up session Chad GPT users wrote without AI; their brain activity did not return to baseline, indicating persistent cognitive debt after tool removal.
7. A Carnegie Mellon and Microsoft study (2025, peer‑reviewed at CHI) analyzed 936 real‑world AI‑assisted tasks performed by 319 knowledge workers.
8. Workers who expressed higher confidence in the AI’s ability applied less critical thinking; those who trusted their own expertise evaluated AI output more carefully.
9. The study warned that long‑term overreliance on AI diminishes independent problem‑solving skills.
10. A larger study of 666 participants found a strong negative correlation between frequent AI use and critical‑thinking scores.
11. Cognitive offloading—the tendency to let digital tools do the thinking—mediated this relationship, with an even stronger correlation between offloading and critical‑thinking decline.
12. Shen and Tamkin’s 2026 preprint on software developers learning a new coding library found that those who fully delegated to AI produced working code but scored 17% lower on conceptual quizzes than peers who did not delegate.
13. A study claiming Chad GPT improves learning performance was retracted last month.
14. A 2024 PNAS study reported that students who critically engaged with AI (asking questions, editing ideas, revising thoughtfully) performed better and reported less mental fatigue than those who merely copied AI output.
15. The MIT study’s participants were adults aged 18‑39; the Carnegie Mellon study surveyed adult knowledge workers; cognitive‑offloading research measured working professionals.
16. Research indicates that when cognitive abilities are not exercised, they atrophy.
17. In a Minneapolis high‑school AP literature class where all technology was banned, student confidence in reading ability rose from 46% in September to 95% in February.
18. A study of college phone bans found that struggling students’ grades improved roughly twice the average effect, with first‑year students benefiting most.
19. Psychology Today analysis states that adults who offload thinking to AI lose previously built capacity, whereas children may never develop that capacity if they rely on AI from the start.
20. A Yale student told CNN that classroom discussions had become flat and predictable, with peers using AI to answer professors’ questions, leading to linguistic and cognitive homogenization.
21. A 2017 study found that merely having a smartphone nearby—even face down and turned off—reduced available cognitive capacity.
22. The UK announced a ban on social media for users under 16; 20 US states ban phones during the school day; France, Spain, Greece, Slovenia, and South Korea have introduced or are working on similar restrictions.
23. UK technology minister Liz Scendle flagged AI chatbots as a separate concern, noting children form one‑to‑one relationships with AI systems not designed with child safety in mind.
24. An Education Next article argued that AI did not destroy critical thinking; rather, it exposed pre‑existing weaknesses, citing stagnant NAEP scores and collegiate learning‑assessment findings that pre‑dated Chad GPT.
25. A Rand Corporation survey from March 2026 found that 70% of students worry AI is eroding their critical‑thinking skills, up from 48% to 65% among high‑schoolers in one year.