OpenAI Podcast

Mark Handley, a longtime networking researcher and UCL professor, and Greg Steinbrecher, a GPU-cluster engineer optimizing large-scale training, explain how AI training breaks traditional networks. They discuss why synchronous GPU workloads stall on small faults, how Multipath Reliable Connection sprays traffic and trims packets to avoid failures, and why making this open standard can lower cost and boost reliability for massive GPU fleets.

Sébastien Bubeck, a mathematician-turned-AI researcher, and Ernest Ryu, an applied mathematician who used AI to crack a long-standing optimization problem, discuss AI’s leap in mathematical ability. They talk about models solving deep research problems, automating long research workflows, literature search versus original discovery, and risks of shallow understanding.

Yunyun Wang, product lead on OpenAI’s life sciences models focusing on bio-risk mitigation. Joy Jiao, research lead linking AI to wet labs and systems biology. They discuss building biology-focused models, safeguards and differentiated access, AI-driven lab automation and autonomous labs, and near-term impacts like drug repurposing and assay-driven evaluation.

Jason Wolf, an OpenAI alignment researcher who builds the Model Spec, explains the public framework that defines intended model behavior. He covers how the spec is written and updated, how instruction hierarchies resolve conflicts, handling tricky edge cases like kids’ questions, and how training and transparency shape real-world model behavior.

Karan Singhal, leader of Health AI Research who studies safe, evaluated models for medicine, and Dr. Nate Gross, a physician and health-policy leader focused on improving care. They talk about training and evaluating models for sensitive health questions. They discuss clinician workflows, partnerships with hospitals, real-world trials, data integration, and practical AI tools for patients and clinicians.

Asad Awan, ad lead at OpenAI who balances user trust, privacy, and ad usefulness. He discusses why ads are being tested now. He explains strict separation between model answers and ads. He covers controls, privacy safeguards for sensitive chats, and how ads can help small businesses. He outlines conservative design, personalization toggles, and future conversational ad ideas.

Sarah Friar, OpenAI CFO, and Vinod Khosla, founder of Khosla Ventures, dive into the challenges of AI demand and compute limitations. They discuss transformative impacts in healthcare, emphasizing AI’s role in augmenting medical professionals. Friar details OpenAI's investment strategies, while Khosla predicts significant advancements in robotics. The conversation also touches on the evolution of AI subscriptions and the pressing need for enterprise adaptation, painting an exciting picture of the AI landscape in the near future.

Christina Kim, a research lead at OpenAI focused on user signals and emotional intelligence, and Laurentia Romaniuk, a product manager specializing in model behavior, dive into the nuances of AI personality. They discuss the evolution from GPT-5 to GPT-5.1, highlighting improvements in reasoning and emotional understanding. The duo unpacks the model switcher, user feedback integration, and the balancing act between steerability and safety. Exciting advancements in memory features enhance personalization, promising a more responsive AI experience.

Join Kevin Weil, Head of OpenAI for Science, and Alex Lupsasca, a physicist and research scientist at OpenAI, as they explore the transformative role of AI in scientific discovery. They discuss GPT-5's ability to accelerate research across fields, from complex calculations in physics to literature searches in biology. Alex shares exciting examples, like uncovering black hole symmetries, while Kevin emphasizes the collaborative potential of AI. With predictions for rapid advancements, they envision a future where AI drastically reshapes scientific workflows and breakthroughs.

Ben Goodger, a visionary in browser product design, and Darin Fisher, a seasoned browser engineer, delve into the transformative potential of ChatGPT Atlas. They explore how AI integrates into daily browsing, enabling personalized experiences and automating tedious tasks. Discussion highlights include the evolution of browsers, technical challenges behind Atlas, and the innovative agent mode that allows for smart task execution. They also share real-world examples of how Atlas enhances productivity, making online experiences more efficient and intuitive.