Mark Zuckerberg's podcast interviews: which to watch and why

Best for: Listeners interested in AI for biology, and anyone weighing a nonprofit against a venture model for deep science.

On uListen: summary · chapters · transcript

A relaxed, technical joint interview with Priscilla Chan and Biohub head Alex Rives, in which Elad Gil and Sarah Guo ask investor-style questions about strategy, talent and drug development.

You'll learn:

• Why he thinks data, even more than compute, holds biology models back: “In order to get the data, it's not just like there's some factory somewhere that you can pay to produce the, the data.” (18:21).
• How patient groups willing to try experimental drugs could change the economics of rare disease (36:36).
• Why he thinks the Biohub can attract the AI talent it needs: a strong group of a dozen or two is enough, and the mission draws people (44:08).

Zuckerberg speaks 39% of the words

Best for: AI and biotech people who want specifics on virtual cell models and how the Biohub is organized; parts of his answers repeat on Latent Space (November 6, 2025).

On uListen: summary · chapters · transcript

A warm joint interview with Priscilla Chan, in which two a16z hosts mix venture framing with technical biology; she covers the medicine, he the strategy and AI modeling.

You'll learn:

• How he explains biology's missing tools to engineers: “it's kind of like you're coding without being able to step through the code and debug things” (3:32).
• What VariantFormer learns from cells before and after a CRISPR edit (24:24).
• Why he thinks many organizational problems are fixed by seating teams together: “it's interesting how many organizational questions or problems you can fix just by having two teams sit together.” (36:54).

Zuckerberg speaks 40% of the words

Best for: AI researchers, engineers and funders thinking about AI for science; less useful for questions about Meta's products.

On uListen: summary · chapters · transcript

A technical, friendly joint interview with Priscilla Chan, recorded at the CZI Imaging Institute, with hosts who frame biology in AI-lab terms such as benchmarks and abstraction layers.

You'll learn:

• What he expects models to do long before they could replace the wet lab, if they ever do: “you're going to be able to have models that can help generate hypotheses, and scientists can apply their taste on which ideas or k-kind of suggestions come from this are worth testing” (27:02).
• What curing all disease means in practice: catching problems early, such as a possibly cancerous mutation (38:41).
• Why he thinks the timeline depends more on the pace of AI than on biology (48:42).

Zuckerberg speaks 50% of the words

Best for: Anyone who wants his AI strategy in one sitting: open source, coding agents, AGI timelines, AI companions, glasses, ads and jobs.

On uListen: summary · chapters · transcript

A dense, analytical one-on-one in which Dwarkesh Patel steelmans other labs' views and Zuckerberg thinks out loud at length.

You'll learn:

• Why he thinks leaderboards are easy to game: his team easily tuned a Llama 4 Maverick variant to the top of Chatbot Arena, while the released model had no such tuning and ranks lower (8:18).
• Which kinds of distillation from foreign models he considers safe (53:39).
• Why he expects AI to make culture funnier and weirder (1:09:56).

Zuckerberg speaks 83% of the words

Best for: Readers who want his own account of the moderation reversal and government pressure, plus his martial arts life.

On uListen: summary · chapters · transcript

A loose, nearly three-hour sit-down that opens on Meta's moderation changes and wanders through jiu-jitsu, hunting, VR, AI and Apple, with Joe Rogan steering through long tangents.

You'll learn:

• Why, in his view, classifier precision drives wrongful takedowns: “if you're at, you know, 90% precision, that means one out of ten things that the classifier takes down is not actually problematic.” (32:21).
• Why he thinks the US should defend its tech companies abroad: “the US government should be defending its companies, right? Not be the tip of the spear attacking its companies.” (22:01).
• What fighting gives him that business often doesn't: a fast feedback loop (1:07:32).

Zuckerberg speaks 67% of the words

Best for: Founders and strategists who want Meta's history in his own words and how it shapes his bets on glasses and open source, and his view of Apple as a rival.

On uListen: summary · chapters · transcript

Recorded live in front of an arena crowd at Chase Center, with Ben Gilbert and David Rosenthal trading a thesis about Meta's survival and detours into its history.

You'll learn:

• Why he ranks his post-2016 political misreading far above the IPO as a mistake: “if the IPO was a year-and-a-half mistake, I think that the political miscalculation was a twenty-year mistake.” (45:38).
• How the 2012 HTML5 bet failed and forced a year-long native rewrite (37:04).
• The good-versus-awesome test he took from early Oculus people, and why he calls social media only good (56:03).

Zuckerberg speaks 63% of the words

Best for: Readers who want his longest relaxed case for the metaverse, plus his most detailed answers on feed ranking and moderation.

On uListen: summary · chapters · transcript

A loose, nearly three-hour talk that opens on the new headset Rogan has just tried and later turns to moderation; on-air details, such as his children being six and five here but nine and seven on his January 10, 2025 Rogan visit, suggest it was recorded well before its listed date.

You'll learn:

• Why Facebook ignores angry reactions when deciding what to spread: “but I'm not here to design a service that makes people angry.” (1:25:59).
• How a wrist band could read signals from little-used motor-neuron pathways (43:18).
• Why he thinks Facebook won: “I think, like, kind of care is probably undervalued in terms of determining what, who ends up doing what in the world.” (2:36:01).

Zuckerberg speaks 72% of the words

Best for: Listeners who want his most technical case for open source AI and Meta's compute buildout, with real pushback.

On uListen: summary · chapters · transcript

A dense technical interview in which Dwarkesh Patel pushes back with concrete hypotheticals and mixes in questions about history.

You'll learn:

• How falling behind TikTok on Reels left Meta with a GPU stockpile (5:06).
• Why he expects energy to limit AI scaling before money does: “But I actually think before we hit that, you're gonna run into energy constraints, right?” (28:06).
• Why he can't stop building: “like constitutionally, like, I'm doing something wrong if I'm not building something new.” (58:35).

Zuckerberg speaks 80% of the words

Best for: Readers who want his thinking on AI for biology and his hardware philosophy in a friendly, non-adversarial setting.

On uListen: summary · chapters · transcript

A long, warm conversation in the couple's home: the first hour with Dr. Priscilla Chan on CZI science, the second one on one about social media, VR, glasses and AI personas.

You'll learn:

• Why he treats language models as hypothesis generators in science (31:04).
• His test for the risk of a new technology: “the bar shouldn't be is it possible to do anything bad, it's does this new technology make it easier to do something bad than what people already had?” (1:58:27).
• Why he separates the real room from the physical room (1:35:51).

Zuckerberg speaks 49% of the words

Best for: Readers who want his fullest explanation of Codec Avatars, plus a snapshot of Quest 3, Meta's first AI personas and Llama 2.

On uListen: summary · chapters · transcript

A demo-style conversation recorded inside the metaverse, with both men appearing as photorealistic Codec Avatars from hundreds of miles apart and Lex Fridman openly awestruck.

You'll learn:

• How Codec Avatars send an encoded model of your face instead of video (2:44).
• Why he would like an avatar more expressive than his own face: “I- I always get a lot of, um, critique and- and shit for, um, for- for having like a relatively stiff, uh, expression.” (9:18).
• Why anonymity online weakens good behavior: “if you stiff them, then like, you know, life is a repeat game and that's not gonna work out well for you.” (49:07).

Zuckerberg speaks 75% of the words

Best for: Readers who want his thinking on Llama, AI products and AI risk at the time, plus candid talk on layoffs, org design and his faith.

On uListen: summary · chapters · transcript

A long, friendly in-person talk between jiu-jitsu training partners, moving from his first tournament through Llama, moderation, layoffs and Quest 3 to faith.

You'll learn:

• Why every business will want its own AI rather than a general assistant (45:43).
• How he wants an AI to answer a question about 3D-printing a gun (1:31:20).
• Why he thinks a pure intelligence will struggle to grasp the human condition: “This gets to the question about AI. I don't think that a being is just a mind.” (2:32:39).

Zuckerberg speaks 78% of the words

Best for: Listeners who want him as a manager and a person: culture design, the Sheryl partnership, parenting and faith.

On uListen: summary · chapters · transcript

A relaxed remote interview in which Tim Ferriss mixes in questions from friends and lets him talk at length, including a long walk through Meta's new values.

You'll learn:

• Why a value is only worth writing down if a good company could choose the opposite (39:37).
• Why Meta moved performance reviews from every six months to once a year (43:34).
• What being an introvert means for him: “people kind of think that, that introverts are, like, you know, for, like, don't like other people or something, and that's not true. I just, just get overwhelmed easily.” (1:24:41).

Zuckerberg speaks 79% of the words

Best for: Viewers who want his metaverse product thinking alongside his answers to critics on polarization, teen wellbeing and free speech.

On uListen: summary · chapters · transcript

A long, earnest in-person conversation: a third on the metaverse, then hard questions framed as steel-manning his critics, and a close on death and meaning.

You'll learn:

• Why every controversial content call is negative sum for Meta's brand (57:28).
• Why, citing research, he argues social media is not a major driver of polarization (53:02).
• What he thinks gives life meaning: “human connection is the meaning. And I, I mean, I think that it's a thing that our society probably systematically undervalues.” (1:56:26).

Zuckerberg speaks 73% of the words