Ginkgo’s Protein Design Contest Puts OpenAI Models Against a Top Scientist
Ginkgo Bioworks CEO Jason Kelly organized a three-round protein-design competition between a leading scientist and OpenAI’s latest models. The event took place in Ginkgo’s automated laboratory spanning 15,000 square feet in Boston, with company executives serving as judges. The contest used a technique for growing proteins in test tubes faster than in cells.
Ginkgo Bioworks CEO Jason Kelly arranged a three-round contest focused on protein design. One side was a leading scientist; the other was OpenAI’s latest models. The setting was Ginkgo’s automated Boston laboratory, a 15,000-square-foot facility, with company executives acting as judges. The contest employed a method that produces proteins in test tubes, described as quicker than cell-based growth.
The event places AI models in direct comparison with human expertise in a specialized biotechnology task. It also showcases automated lab infrastructure and alternative protein-production techniques as part of broader efforts to speed up design and testing.
This contest may draw attention to how AI tools could complement or challenge expert scientists in biotechnology. Researchers and lab workers might see automation and test-tube protein production as ways to shorten experimental cycles, potentially changing workflows and skill demands. Companies developing AI and lab automation could face greater scrutiny over reliability, safety, and attribution. The public may benefit if faster protein design leads to useful applications, though such outcomes remain uncertain.