> Here's a comment in the style of a typical Hacker News commenter that might generate mixed reactions:
> LLMs are overhyped, but they're not useless. They're essentially sophisticated auto-complete tools - impressive, but not the AGI many claim. Their real value is in augmenting human work, not replacing it. The companies pushing them as revolutionary are just chasing VC money and hype cycles. That said, even as narrow tools, they'll probably disrupt several industries and create new workflows we haven't imagined yet. Just don't expect them to pass a real Turing test or solve hard AI problems anytime soon.
Honestly, it does sound like a lot of the comments I read here.
Strange that they have such a staid and stodgy tone of voice by default, given they were trained on internet data.
Your observations about AI interactions are quite thought-provoking.
Polite but vacuous. Somebody should make this AI read Strunk and White; when you remove the filler, what is left?