GPT-5 and an Automated Laboratory: A Scientific Workflow

SUMMARY

In 30 seconds

In this February 2026 short, Matthew Berman describes a partnership between OpenAI and Ginkgo Bioworks to connect GPT-5 to an automated laboratory. He is enthusiastic about continuous experimentation and possible future scientific benefits, which he presents as potential rather than demonstrated medical outcomes.

1A model connected to a labBerman says OpenAI and Ginkgo Bioworks partnered to connect GPT-5 to a laboratory.
2An experimental loopHe describes the system as proposing experiments, running them and learning from the results.
3Possible benefitsBerman sees potential for continuous operation, lower discovery costs and medical breakthroughs, without claiming those outcomes are already achieved.

Takeaways

The demonstration connects GPT-5 with automated laboratory equipment.
Berman describes a loop of proposing experiments, running them and learning from results.
Round-the-clock research, lower discovery costs and medical advances are potential future benefits.

TopicsAI for science Agents

FULL TEXT

Read the edited text by chapter

Show transcript

Transcript only

AI and the laboratory

Berman introduces the partnership and describes an automated experimental loop. He considers possible scientific benefits.

OpenAI partnered with Ginkgo Bioworks to connect GPT-5 to an automated laboratory. The system can propose experiments, run them and learn from the results within the laboratory, according to the demonstration I describe.

I see this as a major demonstration of AI making decisions about experiments and controlling laboratory processes. It shows a way AI can move from digital answers into physical scientific work.

A laboratory running around the clock with AI has enormous potential. I hope such systems could reduce the cost of scientific discovery and support medical breakthroughs, although those benefits are possibilities rather than results established by this short.