Summary
OpenAI co-founder and president Greg Brockman joins Elad Gil and Sarah Guo on the No Priors podcast to discuss a busy week for the company, including the rollout of GPT-6 to over a billion people and a new set of mathematics results. Brockman argues that AI can now generate genuinely new knowledge, pointing to the resolution of century-old math problems, while acknowledging that models are not yet producing entirely new scientific theories.
The conversation extends from mathematics into biology and drug discovery. Brockman explores how regulators might need to rethink clinical trial expectations if AI-driven predictions prove more reliable than traditional trials, drawing a comparison to self-driving cars that become far safer than human drivers. He notes that regulation differs widely between countries such as Australia and the US, and that some trials are already moving to friendlier jurisdictions.
Brockman also makes the case that cybersecurity spending should eventually vastly outstrip software spending, because cheap software makes reliability and security the scarce qualities. He describes a future in which humans stay firmly in control and AI amplifies human agency, and he discusses what recent releases like Astra and DevDay mean for how OpenAI works internally. The episode is a useful look at how one of the industry’s most influential leaders thinks about science, safety, and the economics of software.
📺 Source: No Priors: AI, Machine Learning, Tech, & Startups · Published October 15, 2026
🏷️ Format: Interview
