On: Why do AI chatbots have such a hard time admitting...

[curiousObject]

The RLHF sessions should incentivize responses like “I don’t know” and “I’m not sure”.

But they do not, because that will not attract investment.

[Workaccount2 → curiousObject]

A low/no hallucination model would probably be the closet thing to a literal money printer in tech right now.

[sharemywin → Workaccount2]

the thing is aren't hallucinations and creativity the two sides of the same coin?

[Workaccount2 → sharemywin]

Right, but that is a lever (temperature) that can be moved in real time, even with current models.

[tlb → sharemywin]

There'll be demand for both kinds of models, imaginative and dependable.

[sharemywin → tlb]

I also see RAG playing a role in the dependable fact based side.

the imaginative side might benefit from it too. think like thought joggers or some kinds of save prompts that could be iterated through to generate more robust ideas.