Immunologist Derya Unutmaz of The Jackson Laboratory is using OpenAI's Codex to build tools that analyze and simulate immune cells, bypassing the traditional bottleneck of dedicated software engineers in research labs.
The conversation gets specific at the 13-minute mark, where Unutmaz walks through what simulating immune cell behavior actually requires computationally, and why existing biology software falls short. That section alone reframes what 'AI for science' means in practice versus in press releases.
The final segment, starting at 29:48, addresses where this is heading: AI systems that let scientists run experiments in silico before touching a pipette. Whether that vision holds up under scrutiny is exactly why the full conversation is worth watching.
[WATCH ON YOUTUBE →]