I agree with that to an extent, but how far should the AI model developers go with that? Like if I ask for advice on, let's say, making custom chef's knives then should the AI give me advice not to stab people? Who decides where to draw the line?

▲

podgietaru 6 days ago | parent | next [-]

Further than they went. Google search results hide advice on how to commit suicide, and point towards more helpful things.

He was talking EXPLICITLY about killing himself.

▲

kelnos 6 days ago | parent | prev | next [-]

We should all get to decide, collectively. That's how society works, even if imperfectly.

Someone died who didn't have to. I don't think it's specifically OpenAI's or ChatGPT's fault that he died, but they could have done more to direct him toward getting help, and could have stopped answering questions about how to commit suicide.

	▲	blackqueeriroh 6 days ago \| parent [-]
		How would we decide, collectively? Because currently, that’s what we have done. We have elected the people currently regulating (or not regulating) AI.

▲

etchalon 6 days ago | parent | prev [-]

I think we can all agree that, wherever it is drawn right now, it is not drawn correctly.