Greg Brockman on Astra and the Future of OpenAI

Greg Brockman outlines a product direction in which models do more than answer questions: they could help generate new knowledge, act across applications, and bring together complex information. He connects that ambition to OpenAI’s claimed Navier–Stokes result, Astra computer control, agentic ChatGPT use, and healthcare products.

Announcements highlighted in the interview

• OpenAI claims a result on the Navier–Stokes problem, described as a counterexample or proof of a singularity under certain conditions.
• Astra is intended to control applications from a user’s goal; Brockman cites 3D creation, modelling physical locations, and part design.
• Agentic use is growing alongside conversational ChatGPT, which he says has more than one billion weekly users.
• Capability discovery is becoming a product problem: the AI should suggest the connectors, permissions, and steps needed to complete a task.

Creation and healthcare

For images, Brockman emphasizes quality, editing control, speed, and variety, with uses from presentations to marketing. In healthcare, OpenAI is targeting consumers, clinicians, and hospitals, including an Epic integration; he mentions medical information access, information flow among caregivers, and clinical-trial recruitment.

The broader signal

The stated direction is a more unified experience spanning chat, agents, voice, images, code, and computer control. Brockman argues for iterative deployment: Operator exposed speed and reliability limits before Astra reached a new threshold.

Source

• Chaîne: TBPN
• Vidéo source: https://www.youtube.com/watch?v=qNxuxwb9ksA