LLMs are hypothesis generators, not proof generators. Hallucination means unwant

LLMs are hypothesis generators, not proof generators.

Hallucination means unwanted hypotheses (imagination). But we do often want hypotheses and imagination. We just want to know the difference.

Our organization produces a governance layer that converts LLMs to proof generators. But all that means is that we end hallucination, and state what’s misleading, lying, false, undecidable, possible, untestifiable, unethical or immoral, and what you’re liable for.

We’ve found that ideation from a proof is safe, so we suggest means of correction or cooperation after we have produced that proof.

Net is that you and the LLM producers are asking too much from LLMs. We explain why and what to do about it.

Cheers
CD

http://
runcible.com

(cc:
@BrianRoemmele
)