Hacker News Reader

force the user to decide in the end

And that's...bad?

Not entirely.

I think, it's a bit much to call this "no hallucinations".

Technically true, but in practice you could still choose the wrong result or the probabilities can be off.

That seems like the only possible way to eliminate hallucination, short of a model that is never wrong.

Wait til you hear about how digital circuits work at die level