I understand the skepticism towards Med-Palm2. The only thing they said about its performance was a vague graph with unlabelled axes showing it beating 'expert level'. It's not clear why we should believe them when IBM was making basically identical claims with the same level of evidence a few years.
I certainly hope to be proved wrong, and that medical AI is here at last.
I certainly hope to be proved wrong, and that medical AI is here at last.