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.
Don't just listen to podcasts. Learn from them with transcripts, summaries, and chapters for every episode. Skim, search, and bookmark insights. Learn more
Tejal Patwardhan, research lead at OpenAI, explains how their frontier evals team is creating new ways to measure AI progress as traditional benchmarks become saturated. She delves into the evolution of evaluations, the challenges of multimodal and real-world tasks, and the importance of adapting testing methodologies to accurately capture model capabilities. The discussion also covers the immense, often underestimated, impact of AI on work and various industries, emphasizing the need for thoughtful development and responsible integration.
Last month AI found something mathematicians had missed for decades. Reasoning researchers Alexander Wei, Hongxun Wu, and Lijie Chen join the podcast to discuss how a general-purpose model helped disprove an 80-year-old conjecture from famed mathematician Paul Erdős. They walk through the moment the result started looking real, what it took to verify the proof, and what’s happened since sharing the discovery with the world. They also explore what this means for the future of math and for researc...
This episode delves into the significant advancements of OpenAI's ImageGen 2.0, highlighting its rapid adoption with over 1.5 billion images generated weekly. Product lead Adele Li and researcher Kenji Hata discuss breakthroughs in photorealism, accurate text rendering, and multilingual capabilities. They explore diverse use cases from personal creative expression (like imperfect "MS Paint" styles) to professional tools for education, marketing, and design, and hint at future "creative agent" functionalities, emphasizing ImageGen's role in an "AI Renaissance" alongside tools like Codex.
Training frontier models isn’t as simple as adding more GPUs—one small problem and the whole coordinated dance falls apart. OpenAI’s Mark Handley and Greg Steinbrecher discuss how a new supercomputer network design, used to train some of the company’s latest models, keeps the whole system moving in lockstep, even with record numbers of GPUs. They break down Multipath Reliable Connection, a new protocol OpenAI developed with AMD, Broadcom, Intel, Microsoft, and Nvidia, and why they’re making it a...
This episode explores the "miraculous" advancement of AI in mathematics, highlighting its surprising ability to solve complex problems and even make original discoveries. Researchers discuss how models went from basic math to research-level capabilities, the concept of an "automated researcher" working over longer timelines, and the profound implications for all scientific fields. While AI accelerates discovery and makes math more accessible, the discussion also covers the critical need for continued human expertise and deep understanding to guide these powerful tools responsibly.
What does it take to build AI systems that can actually help scientists? Research lead Joy Jiao and product lead Yunyun Wang discuss how OpenAI is developing models for life sciences and what responsible deployment means in a field with real biosecurity stakes. They explore how AI is already improving research workflows and where it could lead in drug discovery and more autonomous labs — including why a future with less pipetting sounds pretty good to most scientists. Chapters 0:39 Introducing t...
Jason Wolfe from OpenAI's alignment team explains the Model Spec, a crucial framework guiding AI behavior and ensuring alignment with human values. The discussion covers its practical implementation, including the 'chain of command' for resolving instruction conflicts and the iterative process of evolving the spec based on real-world use and new model capabilities. It also compares the Model Spec to Anthropic's Constitution AI and considers its future role in AI development and user interaction.
Dr. Nate Gross and Karan Singhal from OpenAI discuss their strategy for integrating AI into healthcare, emphasizing rigorous training with physicians, ensuring data security, and tailoring responses to individual contexts. They explore how AI can alleviate clinician burden, improve patient access to information, and overcome systemic silos to ultimately deliver more proactive and personalized care.
How should advertising work in an AI product? Asad Awan, one of the ad leads at OpenAI, walks through how the company is approaching this decision and why it’s testing ads in ChatGPT at all. He explains how ads are built to stay separate from the model response, keep conversations with ChatGPT private from advertisers, and give people control over their experience. Chapters 00:00:29 — Mission and principles 00:04:01 — Separation between ads and answers 00:07:31 — Who will see ads 00:08:52 — Inte...
OpenAI CFO Sarah Friar and Khosla Ventures founder Vinod Khosla discuss the state of AI, emphasizing that demand for compute is the primary constraint, not a market bubble. They highlight AI's transformative potential in healthcare and enterprise, creating tangible value and driving significant productivity gains. The conversation also explores future developments in agents and robotics, the responsible integration of ads, and the long-term societal impacts of a deflationary economy driven by AI.
This episode explores the advancements in OpenAI's GPT-5.1, highlighting the shift to all models being reasoning models capable of deeper thought. Researchers discuss the nuanced approach to building "personality" and emotional intelligence (EQ) in AI, driven by extensive user feedback. They delve into balancing user freedom with safety, enhancing steerability and creativity, and the transformative role of features like memory in creating a more personalized and intuitive user experience.
AI is beginning to change how science gets done. Head of OpenAI for Science Kevin Weil and OpenAI research scientist Alex Lupsasca talk about the early signs of acceleration researchers are seeing with GPT-5—from surfacing literature across fields and languages, to speeding up complex calculations, to designing follow-up experiments. They unpack what’s possible today, what doesn’t work yet, and why the next few years could reshape the trajectory of scientific progress across physics, math, biolo...
The episode explores ChatGPT Atlas, an innovative browser from OpenAI designed to leverage AI for a more intuitive web experience. Ben Goodger and Darin Fisher detail how Atlas integrates ChatGPT to act as an agent, automating tasks, personalizing content, and simplifying complex web interactions. They discuss the architectural decisions behind building on Chromium, the future of agentic internet traffic, and the profound impact AI will have on web accessibility and user productivity.
Sam Altman and Greg Brockman of OpenAI, alongside Hock Tan and Charlie Kawwas of Broadcom, detail their new partnership to develop custom silicon and full systems. This collaboration aims to dramatically scale AI infrastructure, including deploying 10 gigawatts of computing power, to meet the world's increasing demand for advanced intelligence. They discuss optimizing for specific workloads, the long-term journey towards AGI, and the vision of making compute abundant and accessible for everyone.
The OpenAI podcast features startups like SchoolAI, which uses AI tutors to empower educators and students, and Jam.dev, which introduced "Please Fix" for codeless website edits, highlighting a new era of "read-write-think" web experiences. Abridge discusses its AI platform for doctor-patient conversations, reducing clerical burden and improving care, while Cursor shares its evolution from code completion to autonomous coding agents, democratizing software development. The episode emphasizes the profound impact of AI on various industries, the importance of user-centric design, and the exciting future of AI-driven innovation.
OpenAI co-founder Greg Brockman and Codex engineering lead Thibault Sottiaux discuss the journey of Codex from early AI code generation to advanced GPT-5 Codex agents. They highlight the importance of 'harnesses,' agentic coding, and breakthroughs in code review, envisioning AI's profound transformation of software development by 2030. The discussion covers solving novel problems, ensuring security, and the future of compute scarcity.
OpenAI Chief Scientist Jakub Pachocki and researcher Szymon Sidor delve into the journey towards Artificial General Intelligence (AGI), sharing their personal paths from high school in Poland to leading AI research. They discuss breakthroughs in AI's reasoning capabilities, its potential to automate scientific discovery, and the limitations of current benchmarks. The episode also touches on the practical implications of AGI for everyday users and offers advice to students interested in the field.
This episode features OpenAI's Head of Education, Leah Belsky, alongside college students, exploring AI's impact on learning. They delve into how AI equalizes access to tutoring, builds student confidence, and redefines necessary workforce skills. Discussions also cover the evolution of academic policies, the "brain rot" debate, and the shift from passive information consumption to active, personalized learning with tools like ChatGPT's Study Mode.
OpenAI's COO Brad Lightcap and Chief Economist Ronnie Chatterji discuss AI's rapid evolution and its profound effects on the global economy. They explore how AI enhances productivity in areas like software development and scientific research, enables small businesses, and democratizes access to intelligence in emerging markets. The conversation also highlights the growing importance of soft skills and adaptability in the AI era, stressing education's pivotal role in preparing for future work.
OpenAI's Nick Turley and Mark Chen discuss the unexpected viral success of ChatGPT, from its naming origin to the technical and cultural shifts needed to scale it. They delve into debates around model behavior, bias, and safety, highlighting the move towards iterative deployment and user freedom. The conversation also explores the breakthrough of ImageGen, the rise of agentic coding, and the essential skills needed for an AI-driven future, such as curiosity, agency, and adaptability.
Sam Altman joins the OpenAI Podcast to explore the evolution of AI, defining AGI and superintelligence. He shares insights on using ChatGPT in daily life, its potential to accelerate scientific discovery, and the massive infrastructure needs of projects like Stargate. The discussion also covers user privacy, monetization challenges, the nuances of human-AI interaction, and OpenAI's future hardware plans, offering a glimpse into the strategic direction of AI development.
Welcome to the new OpenAI podcast, hosted by Andrew Main. This podcast offers a glimpse behind the scenes at OpenAI, featuring conversations with people working with and at the organization. The first episode previews a discussion with CEO Sam Altman, covering topics like Stargate, using ChatGPT as a parent, and the future of AGI.
Jun 13, 2025•1 min
For the best experience, listen in Metacast app for iOS or Android
