Sam Altman's podcast appearances: which to watch and why

Best for: Founders who want his operating philosophy, and anyone following his views on safety and concentrated power.

On uListen: summary · chapters · transcript

A loose, warm talk in which David Senra, with no question list, draws on founder biographies; the recording starts mid-conversation.

You'll learn:

• Why managing research is, to him, the closest thing to startup investing (11:26).
• How Peter Thiel's advice made OpenAI go all in on ChatGPT (56:34).
• What writing letters to his newborn son taught him: “the mindset of writing to your kid. Like, you really can't hide behind anything.” (1:17:32).

Altman speaks 64% of the words

Best for: Investors and founders who want compute economics, moats and Codex, plus his mid-2026 view on AGI, jobs and safety.

On uListen: summary · chapters · transcript

A relaxed long-form interview in which Patrick O'Shaughnessy draws on Altman's own posts, starting with a recent one, and ranges from compute economics to parenting.

You'll learn:

• When he expects robotics' ChatGPT moment: “I would say we get the ChatGPT moment for robotics in the next, like, two or three years.” (35:40).
• The question he thinks gets too little attention: how to avoid cognitive atrophy (49:12).
• His idea of a slider for how much compute a personal AI spends thinking while you sleep (29:55).

Altman speaks 71% of the words

Best for: Students and founders who want the ChatGPT origin story and his current views on AI as a utility, education and power.

On uListen: summary · chapters · transcript

A classroom fireside chat in which an unnamed host, who took Altman's 2014 startup class, asks long questions and then reads students' questions aloud.

You'll learn:

• His theory-free case for pushing what already works to a scale nobody has tried (7:15).
• Why early electricity companies sold light at night, and his admission: “I don't know what our equivalent of we're selling you light at night is going to be.” (21:54).
• Why he thinks every frontier lab must largely become an inference company (25:28).

Altman speaks 73% of the words

Best for: Readers who want his compute thesis in short form. It is promotional, with the partner's executives present, and says little about safety or society.

On uListen: summary · chapters · transcript

A short announcement episode on OpenAI's own channel: Andrew Mayne hosts Altman, Greg Brockman and Broadcom's Hock Tan and Charlie Kawwas, and the others carry most of the technical detail.

You'll learn:

• Why a custom inference chip grew into whole custom racks: “These things have gotten so complex, you need the whole thing.” (0:47).
• The metric he says matters: “what we want is the most intelligence we can get out of each unit of energy.” (17:26).
• His claim that about 2 gigawatts already serves roughly ten percent of the world on ChatGPT (21:27).

Altman speaks 24% of the words

Best for: Readers who want his late-2025 business strategy (vertical stack, infrastructure, Sora, ads and copyright) for an investor audience.

On uListen: summary · chapters · transcript

A friendly two-host conversation: Erik Torenberg works through strategy questions while Ben Horowitz adds operator and policy follow-ups, often riffing at length.

You'll learn:

• Why he bets Sora matters for AGI: “But I would bet that if we can build really great world models, that'll be much more important to AGI than people think.” (5:17).
• That many Sora rights holders worry their characters won't be generated often enough (29:55).
• Why, in his view, price will decide how fast nuclear power arrives (36:46).

Altman speaks 61% of the words

Best for: Young people, especially in India, deciding what to study or build, and founders building on foundation models.

On uListen: summary · chapters · transcript

A short one-on-one recorded the day before the GPT-5 launch, in which Nikhil Kamath asks practical, India-focused questions and presses for specifics.

You'll learn:

• The hard skill he would learn now: “And right now, learning how to use AI tools is probably the most important, specific, hard skill to learn.” (9:14).
• His test for which AI startups survive better models (31:58).
• Why he says no one running things has a plan: “now that I'm supposed to be the adult in the room, I will tell you, I think no one has a plan. No one really has it all working smoothly.” (15:17).

Altman speaks 75% of the words

Best for: Listeners who want a short mid-2025 snapshot: GPT-5, privacy and ads, sycophancy, Stargate and his comments on Elon Musk.

On uListen: summary · chapters · transcript

The first episode of OpenAI's own podcast, on its own channel, hosted by former OpenAI engineer Andrew Mayne, who moves briskly through news hooks.

You'll learn:

• His bar for superintelligence: “if we had a system that was capable of either doing autonomous discovery of new science or greatly increasing the capability of people using the tool to discover new science” (5:19).
• What he saw at the first Stargate site, in Abilene (25:30).
• How he leaned on ChatGPT in his baby's first weeks (0:50).

Altman speaks 63% of the words

Best for: Readers who want concrete 5 to 10 year predictions and an unusually personal look at the job.

On uListen: summary · chapters · transcript

A loose, teasing talk with his brother Jack Altman, moving from 5 to 10 year forecasts to OpenAI's strategy to how the job feels.

You'll learn:

• His main worry, which isn't technical: “that, like, we're gonna solve the superintelligence, but maybe society still sucks.” (12:28).
• What he makes of Meta's $100 million signing bonus offers to OpenAI staff (22:35).
• What OpenAI was meant to be: “It was supposed to be my retirement job. I was supposed to, like, run a little research lab.” (32:42).

Altman speaks 66% of the words

Best for: Readers who want his thinking on meaning, connection and status in an AI world, plus a short account of the firing.

On uListen: summary · chapters · transcript

A research-driven talk in which Adam Grant brings social science studies for Altman to react to and runs a lightning round before closing on ethics and regulation; on air Altman calls o1 the most powerful model he knows of and says he is expecting a child soon.

You'll learn:

• Why he reads our preference for human work as a very good sign (14:38).
• The sadness he would feel if AI took over creative problem-solving, and why he expects people to adapt (9:51).
• Why he is relaxed about leaning on tools: “You know, I can't spell complicated words anymore because I just trust that autocorrect will save me. I feel fine about that.” (23:14).

Altman speaks 57% of the words

Best for: Founders, developers and investors building on OpenAI who want guidance on agents and pricing, plus his leadership lessons.

On uListen: summary · chapters · transcript

A live on-stage interview at OpenAI DevDay in which Harry Stebbings mixes audience questions with his own and ends with a rapid-fire round.

You'll learn:

• His off-the-cuff definition of an agent: “something that I can give a long duration task to and provide minimal supervision during execution for.” (9:48).
• Why he finds the transistor the least bad analogy for AI (31:24).
• A rare admission: “I think that product is a weakness of mine in general.” (36:10).

Altman speaks 76% of the words

Best for: Listeners who want Altman relaxed and personal, with his ideas on UBI, slow takeoff, the pause and global rules.

On uListen: summary · chapters · transcript

A freewheeling conversation in which Joe Rogan does more than half the talking, with most of the AI substance near the start and end; on air GPT-4 is the current model and Rogan says he is doing Sober October, so it was likely recorded months before it was published.

You'll learn:

• Why he would give everyone a tradable slice of the AGI system, not just a check (9:09).
• Why he calls a pause naive at best (2:27:43).
• His account of psychedelic therapy, which he counts among the most important things in his life: “I traveled to another country for, like, a week, did a few things, came back. A totally different person.” (1:45:10).

Altman speaks 42% of the words

Best for: Listeners who want his mid-2024 thinking on assistants, copyright and regulation, plus his own account of the firing.

On uListen: summary · chapters · transcript

A banter-heavy roundtable in which four hosts question him and push back; he leaves well before the end, and the hosts finish on their own with the week's news.

You'll learn:

• Why he wants an assistant that acts like a senior employee who pushes back, not an alter ego (20:39).
• Why he thinks the copyright fight will move from training data to inference time (35:32).
• His regret about equity: “Yeah, I wish I had taken, I wish I had taken equity so I never had to answer this question.” (59:18).

Altman speaks 37% of the words

Best for: Founders and investors deciding what to build on OpenAI's models, and anyone curious how he splits work with his COO.

On uListen: summary · chapters · transcript

A brisk joint interview with OpenAI COO Brad Lightcap, recorded in London at an OpenAI enterprise event and ending with a quick-fire round.

You'll learn:

• How he sees his own role: “I think AGI will be the most important thing I ever touch, but this is not my natural fit.” (15:32).
• Why he expects model makers to shake out to about a dozen, like early carmakers (19:54).
• What the job has cost him: “But I have basically, like, run out of time for real life.” (49:43).

Altman speaks 57% of the words

Best for: Readers who want his account of the board crisis and the Musk dispute, plus a tour of Sora, GPT-4's limits, compute and AGI.

On uListen: summary · chapters · transcript

A long, relaxed one-on-one, his second on the show, in which Lex Fridman opens with the board crisis and the Musk lawsuit and presses when Altman deflects.

You'll learn:

• What he values most in leaders: “the thing I really value in leaders is how people act on a boring Tuesday at 9:46 in the morning.” (17:27).
• His version of why Elon Musk left OpenAI (27:33).
• Why he thinks compute demand depends on price, like energy (1:10:52).

Altman speaks 63% of the words

Best for: Anyone who wants his early-2023 worldview in depth: GPT-4 basics, bias, AGI timelines, OpenAI's capped-profit structure, jobs and UBI.

On uListen: summary · chapters · transcript

A long, earnest sit-down recorded six days after the GPT-4 launch, in which Lex Fridman asks broad philosophical questions and Altman often turns them back on him.

You'll learn:

• How RLHF works in its simplest form (6:45).
• Why he wants users treated like adults: “One small thing that really bothers me about our current thing, and we'll get this better, is I don't like the feeling of being scolded by a computer.” (41:04).
• Ilya Sutskever's test for whether a model is conscious (1:05:11).

Altman speaks 55% of the words