Conversation

Notices

1. Embed this notice
   Christine Lemmer-Webber (cwebber@social.coop)'s status on Monday, 16-Jun-2025 21:18:24 JST Christine Lemmer-Webber

   Your Brain on ChatGPT: Accumulation
   of Cognitive Debt when Using an AI
   Assistant for Essay Writing Task https://arxiv.org/pdf/2506.08872

   Fascinating paper. It's long, 200 pages; you can read the intro and the Discussion section and get much out of it already though, but they do their homework.

   Effects of LLM usage on memory retention and task execution:

   - LLM users do get judged well by human and LLM judges for their essays
   - Can't remember them
   - Low cognitive activation
   - Essays are mostly the same

   • GreenSkyOverMe (Monika) repeated this.
   • Embed this notice
     Christine Lemmer-Webber (cwebber@social.coop)'s status on Monday, 16-Jun-2025 21:20:02 JST Christine Lemmer-Webber

     in reply to

     • Hailey

     HT to @hailey for linking this one; it's really good to see something this robust.

     I think many of us have suspected that this is the kind of effect one would have when using LLMs heavily: you don't learn, you don't remember, you become a conduit for the LLM tooling. But here's some clear information.

     Comparison to search engine usage is also done in the paper.

     Cadu Silva :verifiedcat: repeated this.
   • Embed this notice
     Crovanian (CamstonIsland) (crovanian@mastodon.social)'s status on Monday, 16-Jun-2025 23:58:34 JST Crovanian (CamstonIsland)

     in reply to

     @cwebber 200 pages, geez. Crok, can you summarize this?

   • Embed this notice
     djuber (djuber@fosstodon.org)'s status on Tuesday, 17-Jun-2025 00:55:49 JST djuber

     in reply to

     @cwebber the "How to read this paper" guide is a welcome addition at the end of the abstract, as though they expected a lot of people to want to read their results, and sympathized with them. I think that's the first time I've seen this, and wish every author was so clear in their anticipation of their readers' goals.

     Christine Lemmer-Webber repeated this.
   • Embed this notice
     Indieterminacy (indieterminacy@social.coop)'s status on Tuesday, 17-Jun-2025 04:27:28 JST Indieterminacy

     in reply to

     • Hailey

     @cwebber Out of curiosity, as somebody who got into advanced typing techniques (I recently heard your talk on that peeling injury), have you had any thoughts on typing; recall; and throughput?

     For instance, I feel ambiguous when I valorize typing to the point where its borderline repetitive. Im unsure whether Ive succeeded (enough to think through process) or merely am missing one or more procedures to then hit real artisanal strides. @hailey

   • Embed this notice
     Christine Lemmer-Webber (cwebber@social.coop)'s status on Tuesday, 17-Jun-2025 04:36:36 JST Christine Lemmer-Webber

     in reply to

     • Hailey
     • Indieterminacy

     @indieterminacy @hailey Ah, well! LLMs for *dictation* are a great idea

   • Embed this notice
     Indieterminacy (indieterminacy@social.coop)'s status on Tuesday, 17-Jun-2025 04:44:05 JST Indieterminacy

     in reply to

     • Hailey

     @cwebber Sure, provided that classic usability isnt dropped (this has affected many communities).

     Ive been intrigued by Jean Louis' enthusiasm for LLM dictation is worth noting:
     https://list.orgmode.org/Z9h8IIiJsYyZXTeD@false/

     Not only for Jean's specialism on #pkm but additionally his historic emphasizing libresoft philosophy.

     (Itching for the chance to scat inside Emacs should I ever get the time).
     @hailey

   • Embed this notice
     nataliyakosmyna (nataliyakosmyna@mastodon.social)'s status on Tuesday, 17-Jun-2025 08:53:11 JST nataliyakosmyna

     in reply to

     • djuber

     @djuber @cwebber we see you !

   • Embed this notice
     Christine Lemmer-Webber (cwebber@social.coop)'s status on Tuesday, 17-Jun-2025 08:53:48 JST Christine Lemmer-Webber

     in reply to

     • djuber
     • nataliyakosmyna

     @nataliyakosmyna @djuber Oh I absolutely love it whenever I'm talking about a paper and one of the authors respond! You all did an incredible job! Your whole team should feel proud!

   • Embed this notice
     nataliyakosmyna (nataliyakosmyna@mastodon.social)'s status on Tuesday, 17-Jun-2025 08:54:45 JST nataliyakosmyna

     in reply to

     • Hailey

     @cwebber @hailey thank you for mentioning the Search Engine group. I feel seen :-) (I’m the lead author of the paper)

   • Embed this notice
     Christine Lemmer-Webber (cwebber@social.coop)'s status on Tuesday, 17-Jun-2025 08:55:13 JST Christine Lemmer-Webber

     in reply to

     • Hailey
     • nataliyakosmyna

     @nataliyakosmyna @hailey You all did an awesome job! 💜 🏆 🌟

   • Embed this notice
     Christine Lemmer-Webber (cwebber@social.coop)'s status on Tuesday, 17-Jun-2025 09:01:50 JST Christine Lemmer-Webber

     in reply to

     • nataliyakosmyna

     The Your Brain on ChatGPT lead author @nataliyakosmyna wrote a nice summary thread about the paper here, on the fediverse! You should check it out and boost the hecc out of it! https://mastodon.social/@nataliyakosmyna/114695325379387766

   • Embed this notice
     nataliyakosmyna (nataliyakosmyna@mastodon.social)'s status on Tuesday, 17-Jun-2025 09:05:50 JST nataliyakosmyna

     in reply to

     • djuber

     @cwebber @djuber thank you! I do love to hear from the “source” as well! We are tired but proud indeed!

   • Embed this notice
     nataliyakosmyna (nataliyakosmyna@mastodon.social)'s status on Tuesday, 17-Jun-2025 23:56:31 JST nataliyakosmyna

     in reply to

     • Wulfy

     @n_dimension @cwebber it is a preprint indeed! My first one, to be honest (two other papers on arxiv under my profile were not submitted by myself), and I see your points about the preprints, I also prefer the papers to be peer-reviewed first! But I felt like in 6-8 months (best case!), when the paper will undergo the peer reviews, some policymaker will already decide to fully vote in favor of LLM to be integrated in all kindergartens across the country.

   • Embed this notice
     Wulfy (n_dimension@infosec.exchange)'s status on Tuesday, 17-Jun-2025 23:56:32 JST Wulfy

     in reply to

     • nataliyakosmyna

     @cwebber @nataliyakosmyna

     Caveat. It's a preprint.
     Not peer reviewed yet.
     I'm not rejecting it outright like many AI papers for flawed methodology.

     Intuitively I'm sceptical of the brainwave analysis. I have to read the full paper how it was measured.

     One interpretation might be that the LLM users were chill about what they were doing having confidence in their tool.

     THINK HARDER seems like a hard deal to sell.
     Human civilization is driven by less effort, not more.

     Christine Lemmer-Webber repeated this.
   • Embed this notice
     Christine Lemmer-Webber (cwebber@social.coop)'s status on Wednesday, 18-Jun-2025 00:01:33 JST Christine Lemmer-Webber

     in reply to

     • Wulfy
     • nataliyakosmyna

     @nataliyakosmyna I think that's wise. I did read as much of it as I could between other things yesterday, and I was impressed by how thorough it was.

     It's a preprint, sure, but AI whitepapers by AI corporations are being treated as canonical sources with way less rigor to them right now.

     BTW @n_dimension your bio says:

     > "Humans are the larval form of AI" - me

     > "Only thing that stops a bad guy with an AI... Is a good guy with an AI." - me

     so I am guessing that's relevant to your response