I like how many of these contrived test conditions disallow the use of LLMs ... as if that's a realistic expectation in this modern era.
There have been a few comparisons where physicians could use LLMs, and that only brought physicians up to the level of the LLM. I would suspect physician + any resources (including OpenEvidence etc.) would score much higher on this NOHARM ranking, but still wouldn't be the best (and, if time were a factor, possibly also the slowest!).
I like how many of these contrived test conditions disallow the use of LLMs ... as if that's a realistic expectation in this modern era.
There have been a few comparisons where physicians could use LLMs, and that only brought physicians up to the level of the LLM. I would suspect physician + any resources (including OpenEvidence etc.) would score much higher on this NOHARM ranking, but still wouldn't be the best (and, if time were a factor, possibly also the slowest!).