People defer thinking about what correct and incorrect actually looks like for a whole wide scope of scenarios and instead choose to discover through tr

People defer thinking about what correct and incorrect actually looks like for a whole wide scope of scenarios and instead choose to discover through trial and error. LLMs are _still_ terrible at deriving even the simplest of logical entailment. I've had the latest and greatest Claude and GPT derive 'B instead of '(not B) from '(and A (not B)) when 'A and 'B are anything but the simplest of English sentences. I shudder to think what they decide the correct interpretations of a spec written in prose is.