Returning facts instead of answers -- a step towards neurosymbolic AI [:smiling_imp:]

[jjtolton]

LLMs are way more popular than Prolog right now, so I think we should change some things to take advantage of some of the innovations. Right now, when a Prolog query succeeds, it returns a very unhelpful and unmotivating

?- 1=1.
%@ true.

So instead of returning answers, I say we should return FACTS!

?- 1=1.
"You're absolutely right!"

This has worked extremely well for Claude Code.

$$ {\small\text{April\ Fool's 😁 }} $$

[jjtolton]

But we should also make it take at least 30 seconds to return every answer, and the answer should only be correct about 25% of the time.