OpenAI Tested AI on 100 Real Lab Tasks. The Hard Problem Is Knowing When Not to Trust It
OpenAI BioBasicsBench tested 100 wet-lab problems: 66.7% versus 43.8% mean rubric credit. What that means for biotech AI reliability and adoption.
EXPLORE A LITTLE FURTHER
1 stories from the Compound journal.
OpenAI BioBasicsBench tested 100 wet-lab problems: 66.7% versus 43.8% mean rubric credit. What that means for biotech AI reliability and adoption.