4 ms·General-purpose large language models outperform specialized clinical AI tools5 points by benwen 4mo ago