Hosted by Andrew Mayne, The OpenAI Podcast features conversations with the people working at and building with OpenAI. Topics range from what goes into developing frontier AI models and new features, to what users are doing with the technology. It’s a practical look at how AI is made and where it’s going, told by the people closest to the work. Hosted on Acast. See acast.com/privacy (https://acast.com/privacy) for more information.
What this show covers
reasoning modelsAI infrastructuresupercomputer networkingdistributed systemsGPU clustersopen standardsbenchmarksfuture of workAI evaluationmultimodal AIphotorealismAI image generationcreative toolsImages 2.0prompting
What listeners will find here
Why bigger AI training runs need a smarter network, not just more GPUs
This episode explores why training frontier AI models pushes computer networks to their limits and how OpenAI’s new Multipath Reliable Connection protocol is designed to keep massive GPU clusters synchronized. Mark Handley and Greg Steinbrecher explain the bottlenecks, failure modes, and the case for making the approach an open standard.
Why frontier AI tests keep failing as models get smarter
Frontier evals are the focus here, as Tejal Patwardhan and Andrew Mayne explore how to measure progress when older benchmarks become too easy. The episode looks at reasoning, surprising model behavior, multimodal testing, and what future evaluations should actually capture.
Why AI image generation is exploding—and where it’s headed next
Image generation is now happening at massive scale, with over 1.5 billion images created each week in ChatGPT. Adele Li, Kenji Hata, and Andrew Mayne look at how Images 2.0 improved realism, text rendering, and creative control, while exploring the trends and workflows shaping what comes next.
How AI helped topple an 80-year-old Erdős conjecture
This episode examines how a general-purpose reasoning model helped disprove Paul Erdős’s unit distance conjecture. Alexander Wei, Hongxun Wu, and Lijie Chen discuss the proof, how they verified it, and what the result could mean for mathematicians and AI research.
How AI is changing racing without taking humans out of the driver’s seat
Racing is the subject here, with a focus on how AI helps teams turn huge amounts of track, garage, and driver data into better decisions. Joyce Ruffell and Chase Holden explore collaboration, competition, and why smaller teams may benefit most from smarter tools.
Episodes
More like OpenAI Podcast
B2B Podcasting Insights - B2B sales & audience engagement strategies for founders and brands
Podknows Podcasting - B2B Podcasting Experts
Websites Made Simple: How to have a successful website design business
Holly Christie (Website Design Business Mentor)
AI Business Daily
Cast Nexa
Jack Hopkins
Jack Hopkins
The Daily Stoic
Daily Stoic | Backyard Ventures
The a16z Show
Andreessen Horowitz
The Ramsey Show
Ramsey Network
All-In with Chamath, Jason, Sacks & Friedberg
All-In Podcast, LLC
Equity Mates Investing Podcast
Equity Mates Media
The Jordan Harbinger Show
Jordan Harbinger