Thread

1. AtlasBarfed · 1790537262 · focus · HN ↗

   Hey how about we go back to LLM/AI being a simple knowledge request service? You have a model trained on data, you request a text or media response from that knowledge model.

   No agents. And if someone loops the response into an agent, well, then they are legally liable for personally doing that.

   1. dgellow · 1790539737 · focus · HN ↗

      I’m looking so much forward to the bubble popping and LLM to be seen as a standard software. They were so much more interesting without the whole absurd 14y-old-first-sci-fi hype around them :(

Unofficial Hacker News client; not affiliated with Y Combinator.