I think the big mistake is _blindly relying on the results_ - although that problem has been improving dramatically (gpt3.5 hallucinated constantly, I rarely se

I think the big mistake is _blindly relying on the results_ - although that problem has been improving dramatically (gpt3.5 hallucinated constantly, I rarely see a hallucination w/ the latest gpt/claude models)