HackerPapers

arXiv:2506.08872cs.AI

Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task

Nataliya Kosmyna, Eugene Hauptmann, Ye Tong Yuan, Jessica Situ, Xian-Hao Liao, Ashly Vivian Beresnitzky, Iris Braunstein, Pattie Maes

Abstract

This study explores the neural and behavioral consequences of LLM-assisted essay writing. Participants were divided into three groups: LLM, Search Engine, and Brain-only (no tools). Each completed three sessions under the same condition. In a fourth session, LLM users were reassigned to Brain-only group (LLM-to-Brain), and Brain-only users were reassigned to LLM condition (Brain-to-LLM). A total of 54 participants took part in Sessions 1-3, with 18 completing session 4. We used electroencephalography (EEG) to assess cognitive load during essay writing, and analyzed essays using NLP, as well as scoring essays with the help from human teachers and an AI judge. Across groups, NERs, n-gram patterns, and topic ontology showed within-group homogeneity. EEG revealed significant differences in brain connectivity: Brain-only participants exhibited the strongest, most distributed networks; Search Engine users showed moderate engagement; and LLM users displayed the weakest connectivity. Cognitive activity scaled down in relation to external tool use. In session 4, LLM-to-Brain participants showed reduced alpha and beta connectivity, indicating under-engagement. Brain-to-LLM users exhibited higher memory recall and activation of occipito-parietal and prefrontal areas, similar to Search Engine users. Self-reported ownership of essays was the lowest in the LLM group and the highest in the Brain-only group. LLM users also struggled to accurately quote their own work. While LLMs offer immediate convenience, our findings highlight potential cognitive costs. Over four months, LLM users consistently underperformed at neural, linguistic, and behavioral levels. These results raise concerns about the long-term educational implications of LLM reliance and underscore the need for deeper inquiry into AI's role in learning.

arXiv comments
216 pages, 102 figures, 4 tables and appendix
Journal reference
DOI
Subjects
cs.AIArtificial Intelligence

On Hacker News

Linked by 22 identifiable commenters in 28 comments across 24 threads, from Jun 2025 to Jul 2026. Found in 5 HN submissions from Jun 2025 to Jun 2025. Open submission discussions.

Comments linking this paper, per year.

Discussed in comments24 threads

  1. I love LLMs, I hate hype 1 comment

    • The fact that you're even saying this it is probably an admission that you do think it's making you dumber. Most people I know, who are honest with themselves, have admitted to me that they feel like it's making them dumber or "zombifying" them. This is also well studied…

      dakolliView on HNParent context

  2. AI boom risks global financial crash, warn central bankers 1 comment

    • Multiple educators and institutes are grappling with a plague of cheating. This is just a recent article : https://news.ycombinator.com/item?id=48708991 LLMs are found to erode skill > https://arxiv.org/abs/2506.08872 LLMs have largely been a net negative on human education…

      intendedView on HNParent context

  3. Norway imposes near ban on AI in elementary school 2 comments

    • Completely agree, and I think outsourcing critical thinking to AI is a bad idea for learning. Here is one study that researched AI assistance in a writing/learning context: https://arxiv.org/abs/2506.08872. The teachers that have instructed using AI for assistance in our school…

      bendrivView on HNParent context

    • The post from my feed: https://davidepstein.substack.com/p/how-to-use-chatgpt-witho..., with the corresponding article: https://arxiv.org/abs/2506.08872

      bendrivView on HNParent context

  4. Expertise in the age of AI 1 comment

    • This is already studied, people do not retain knowledge when learning with AI. Learning with AI only creates the most mediocre of people, I've witnessed this myself over and over and over again over the last couple of years. Read a book, write, think and you'll be fine. Use LLM…

      dakolliView on HNParent context

  5. Have a Coherent AI Policy 1 comment

    • hm social media maybe you could consider causes cognitive decline similar to what is theorized about over-reliance on an llm. edit: https://arxiv.org/abs/2506.08872

      rsoto2View on HNParent context

  6. AI chatbots could be making you stupider 1 comment

    • I recommend people look at the actual study and think about how representative are the subjects, the tasks involved (SAT essay writing), and the way LLMs are being used. https://arxiv.org/abs/2506.08872 To be concrete, this is taking a task in isolation that LLMs can do much…

      dcreView on HNParent context

  7. AI-assisted cognition endangers human development? 1 comment

    • > When I read pieces like this all I think is, resistance to change is a helluva drug. When I read comments like yours, I’m reminded of (though I’m not comparing you to—I believe you are arguing in good faith) the cryptocurrency shills saying anyone who is against…

      latexrView on HNParent context

  8. AI could be the end of the digital wave, not the next big thing 1 comment

    • It a side effect of using AI. People using AI for tasks (essay writing in the MIT study linked below) showed lower ownership, brain connectivity, and ability to quote their work accurately. > https://arxiv.org/abs/2506.08872 There was a MSFT and Carnegie Mellon study that saw a…

      intendedView on HNParent context

  9. The future of everything is lies, I guess – Part 5: Annoyances 1 comment

    • MIT actually has a paper on how ChatGPT use impacted cognitive skills for essay writing. > Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task > https://arxiv.org/abs/2506.08872 > Cognitive activity scaled down in relation to…

      intendedView on HNParent context

  10. AI assistance when contributing to the Linux kernel 1 comment

    • People have measurably lower levels of ownership and understanding of AI generated code. The people using GenAI reap a major time and cognitive effort savings, but the task of verification is shifted to the maintainer. In essence, we get the output without the matching mental…

      intendedView on HNParent context

  11. OpenAI says its new model GPT-2 is too dangerous to release (2019) 1 comment

    • > it's really hard to sometimes break out of that loop and do manual fixes You're aware of the MIT Media Lab study[0] from last summer regarding LLM usage and eroding critical thinking skills...? [0] Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI…

      logifailView on HNParent context

  12. I'm OK being left behind, thanks 1 comment

    • The kid who showed his work in detail in math class is doing better in life 9/10 times than the kids that only knew how to use a calculator. Now consider how well the people who think you just need to know how to yell at the calculator are going to do? When Maps apps came…

      dakolliView on HNParent context

  13. 747s and coding agents 1 comment

    • Socrates was proven dead wrong by neurobiology. LLMs making you dumber is far from being "disproven" by science. Quite the opposite https://arxiv.org/abs/2506.08872

      cafebabbeView on HNParent context

  14. Vitamin D and Omega-3 have a larger effect on depression than antidepressants 1 comment

    • I don't need to "prove it", because all I have to do is link this: https://arxiv.org/abs/2409.01754 https://arxiv.org/abs/2508.01491 https://aclanthology.org/2025.acl-short.47/ https://arxiv.org/abs/2506.06166 https://en.wikipedia.org/wiki/Wikipedia:Signs_of_AI_writing…

      Der_EinzigeView on HNParent context

  15. Your brain on ChatGPT: Accumulation of cognitive debt when using an AI assistant 2 comments

    • It seems this study has been discussed on HN before, though was recently revised very late December 2025. https://arxiv.org/abs/2506.08872

      misswaterfairyView on HNParent context

    • I'm very impressed. This isn't a paper so much as a monograph. And I'm very inclined to agree with the results of this study, which makes me suspicious. To what journal was this submitted? Where's the peer review? Has anyone gone through the paper…

      xenophonfView on HNParent context

  16. Programming, Evolved: Lessons and Observations 1 comment

    • Traditional engineering involves more than talking to people, teams, and trade offs. Where I live the engineering society here does license software engineers now and enforces the term, "Software Engineer," as a protected term. You can't just call yourself a software engineer.…

      agentultraView on HNParent context

  17. Who's Submitting AI-Tainted Filings in Court? 1 comment
  18. MIT Study Finds AI Use Reprograms the Brain, Leading to Cognitive Decline 1 comment

    • A few things to note. 1. This is arxiv - before publication or peer review. Grain of salt.[0] 2. 18 participants per cohort 3. 54 participants total Given the low N and the likelihood that this is drawn from 18-22 year olds attending MIT, one should expect an uphill battle for…

      tomrodView on HNParent context

  19. My "Are you presuming most people are stupid?" test 2 comments

    • The referenced study is quite the doozy: > We assigned participants to three groups: LLM group, Search Engine group, Brain-only group ... to write an essay. We recruited a total of 54 participants... > We used electroencephalography (EEG) to record participants' brain activity…

      djoldmanView on HNParent context

    • > a billion people are using an app that adds absolutely nothing to their lives each week I like this reasoning. If something is popular it is objectively good. For example 21.7% of adults on earth use tobacco, so it must be good then.…

      jrflowersView on HNParent context

  20. Who is using AI to code? Global diffusion and impact of generative AI 1 comment

    • Strangely enough, I work with AI models as a programmer and do not use any AI tools to code. The creation of "cognitive debt" as AI researchers describe it means that crucial components of my code are offloaded into a black box understanding. https://arxiv.org/abs/2506.08872…

      bdbenton5255View on HNParent context

  21. Your Brain on ChatGPT: Accumulation of Cognitive Debt 1 comment
  22. Your Brain on ChatGPT: Accumulation of Cognitive Debt When Using an AI Assistant 1 comment
  23. Generative AI coding tools and agents do not work for me 2 comments

    • Maybe, but there's also some evidence that AI coding tools aren't making anyone more productive. One study from last year found that there was no increase in developer velocity but a dramatic increase in bugs.[1] Granted, the technology has advanced since this study, but many of…

      lexandstuffView on HNParent context

    • Somehow you think the burden of proof is on his response rather than the AI maxis initial claims, but regardless, here you go. It was measured. https://arxiv.org/abs/2506.08872

      dlock17View on HNParent context

  24. League of Professional System Administrators - Board to Dissolve Organization 1 comment

    • Cloud was a nice promise that was broken all too quickly. This is why cloud repatriation is a growing trend, that and a growingly brittle cloud with outages that are outside anyone's control, and the costs unmanageable. There are also business sectors where they never could…

      trod1234View on HNParent context

Submitted to Hacker News

Connected through shared HN threads, not inferred similarity or a recommendation.