Sure, a system that could systematically and reliably turn LLM output (or other text) into checkable queries like that would be useful.

→ View original post on X — @garymarcus