
@Wildlaw406 @tomweedlaw But given how LLMs are built (large samples and correlation etc), I think it's basically impossible they'll ever be good at areas that repurpose basic terms into terms of art, because their training data is actively harmful to interpretation
English












