• Activity
  • Votes
  • Comments
  • New
  • All activity
  • Showing only topics in ~comp with the tag "llm". Back to normal view / Search all groups
    1. LLM users: how do you feel about it?

      I’ve been using LLMs more lately. For reference, I’m a PhD student (in computer science). I’ve been an LLM sceptic for a long time, and on pretty much all fronts: concern about their output...

      I’ve been using LLMs more lately. For reference, I’m a PhD student (in computer science). I’ve been an LLM sceptic for a long time, and on pretty much all fronts: concern about their output quality, business practices of the companies that run them, data privacy and ethics concerns, and about their negative impact on our cognition and emotions (studies showing experts got worse at their skills after using AI, people, even those aware of how LLMs work, getting emotionally attached to them, and so on).

      Only one of those has changed for me recently - it’s now undeniable that they are extremely useful, and their output can be very high quality (still somewhat task dependent, but their are good at a lot things that are helpful for me). I’m on the $100 ChatGPT subscription now (funded by my research group), and I used it a lot recently, especially with handling admin and details around a paper I recently submitted to a conference, helping with formatting and proof reading extensively. I don’t think I would have been able to get it across the line in time without the use of an LLM.

      Even though my usage is still relatively light - I’m not out here vibe coding or vibe-writing papers - I do feel a bit uncomfortable about it all. And my usage is not that light, else I would be on the $20 tier, but I use Astra enough that the $10 tier is needed. I worry about depending on it, about losing or failing to develop skills I would have otherwise, and about becoming emotionally attached to it. Especially the last point seems extremely unlikely, but I keep reading stories about normal people, smart people, who know what LLMs are and how they work, finding themselves in strange situations, even with these models/products that are designed for work.

      It’s kind of like how I know caffeine is helpful for me and can increase my productivity, but I use it very judicially (else it affects my sleep). And how I don’t touch energy drinks or nicotine, because they have too many risks and adverse affects. But we still don’t really know how LLMs are affecting us cognitively and emotionally, and it does scare me a bit.

      At a conscious level, it’s just a tool, and I’m in control, and I haven’t noticed or experienced any of the issues I’m concerned about, but I’m still, just a little bit, worried. Not wracked with anxiety, but just in the back of my mind, something tells me that all this stuff with AI is wrong. As unscientific as this will sound, that little gut feeling has never led me wrong in the past.

      So I’m curious - other Tilderinos who are using LLMs, or have maybe interacted with them but decided not to use them, or otherwise have a perspective to share: what are thinking and feeling about LLMs?

      1 vote