Hacker News Reader

People have literally tried to hack together an agent on top of GTP-4 and given it an explicit goal of "kill all humans" for the lulz (ChaosGPT). I don't think that particular case is going anywhere, but assuming that nothing roughly GPT-shaped can acquire harmful goals seems naive.

View on HN

Respectfully, you're mixing things again: you have described 2 separate unrelated things:

* A system like chat gpt that can do things a human cannot (find a way to kill all humans that's achievable by a small group).

* A system different to chatgpt that has its own motives and initiative.

Both need to be achieved.

I was responding to:

> No rate improvement in chat GPT will make it "decide to kill all the humans" etc.

All I'm trying to say is that GPT-style architectures not having "agency" shouldn't be very reassuring. How intelligent they can get is a separate question and one on which I am agnostic.

Reply on news.ycombinator.com