Verifiable Environments for AI in Biology — Kenny Workman, LatchBio
Kenny Workman from LatchBio discusses using AI for actual biological research.
Biological experiments generate enormous amounts of data—two to six terabytes per experiment—too much for researchers to analyze manually. LatchBio has developed tools that adapt coding models to biology and built benchmarks to test whether AI actually solves real biological problems. The major challenge is that today's advanced AI models are not yet reliable enough for this work, and biology is harder than many other fields because experiments themselves are often messy and researchers rarely completely agree on answers.
Open on YouTube →
Vibekollen prepared this summary with AI from the original publication. The content belongs to AI Engineer.
