Login
You're viewing the y.diskcat.com public feed.
  • Feb 22, 2026, 5:07 PM

    I hate the current AI landscape as much as the next person. I associate it with excessive energy usage, disinformation and exploitation.

    However, I also have to wonder reading my Mastodon timeline. There seems to be a sentiment that LLMs are inherently both unethical and useless and I just struggle to understand that point of view. Like many things in this world it's probably difficult to make them perfectly ethical, but how could one see no use case at all for the capable models of 2026?

    💬 1🔄 0⭐ 0

Replies

  • Feb 22, 2026, 6:41 PM

    @ju really wanna know what 'capable' models you came across. I'm not even kidding when I say this, most of the models I've used, claude sonnet/opus included, are absolute bollocks in programming, and repeatedly give me code that's extremely buggy, and in particular, just flat-out wouldn't work.

    💬 0🔄 0⭐ 0
  • Feb 22, 2026, 6:50 PM

    @innocentzero I only used opus recently with rust and python. Not saying it is perfect, but I am pretty impressed with its ability to explore code bases or design features.

    Can you think of a reason why they would struggle more with the tasks you give them?

    💬 0🔄 0⭐ 0