Podcasts

AI podcasts are audio shows about artificial intelligence: long form research interviews, daily news roundups, and conversations with the founders, scientists, and investors building the technology. They have become one of the main ways researchers, founders, and engineers keep up with a field that now moves faster than academic publishing, and they are often where new ideas are aired before they reach papers or product launches. The biggest shows routinely host the leaders of OpenAI, Anthropic, and Google DeepMind before those figures speak anywhere else; the Lex Fridman Podcast, the largest, surpassed 5 million YouTube subscribers by 2026. ^[1][19]

This directory covers the most influential AI and machine learning podcasts: shows focused on research, on the business of AI, on policy and safety, and on building with large language models. The list reflects what working AI practitioners recommend most often.

See also: Guides and Books on AI.

What are the most influential AI podcasts?

The table below summarises the most cited AI podcasts, their hosts, launch years, and primary focus.

Scroll sideways for more →

Research and technical podcasts

Lex Fridman Podcast

The Lex Fridman Podcast is the most widely listened to long form AI show. Lex Fridman, a research scientist who has worked at MIT, started the podcast in 2018 under the name The Artificial Intelligence Podcast. He renamed it in 2020 to reflect the broader range of guests, which by then included physicists, historians, and politicians. ^[1][2]

The show is best known for its multi hour conversations with figures such as Ilya Sutskever, Demis Hassabis, Andrej Karpathy, Yann LeCun, Sam Altman, and Dario Amodei. Episodes typically run between two and five hours, with little editing. By 2024 the Boston Globe reported the show had passed 3.6 million subscribers, and the associated YouTube channel surpassed 5 million subscribers with more than 900 million total views by 2026, making it one of the largest tech adjacent podcasts in the world. ^[2][19] The show remained a central venue for AI discussion into 2026: episode 490, "State of AI in 2026," ran a multi hour survey of LLMs, coding agents, scaling laws, China, GPUs, and AGI. ^[20]

The TWIML AI Podcast

Sam Charrington launched This Week in Machine Learning & AI in mid 2016, which makes it one of the oldest dedicated ML shows still running; by 2026 it had released more than 780 episodes. ^[3] The name was shortened to The TWIML AI Podcast as the field grew. Charrington is an industry analyst, and episodes lean toward practitioners working on applied ML, MLOps, and the engineering side of deep learning rather than purely academic conversations. Guests have included researchers from Google DeepMind, OpenAI, Microsoft Research, and dozens of startups.

The Gradient Podcast

The Gradient is a magazine and podcast founded in 2017 by graduate students and researchers at the Stanford AI Lab. The podcast began in 2021, with Daniel Bashir, a machine learning compiler engineer, as the host. ^[8] Episodes are typically long and dense, with researcher guests walking through their papers, methods, and broader views about where the field is going. Topics often include in context learning, mechanistic interpretability, theoretical neuroscience, and AI policy.

Machine Learning Street Talk

Machine Learning Street Talk (MLST) began on 24 April 2020. The original hosts were three YouTubers working in ML: Tim Scarfe, Connor Shorten, and Yannic Kilcher. The show is now mainly run by Tim Scarfe, who has a PhD in machine learning, with frequent appearances from Keith Duggar. ^[12]

MLST is more philosophically inclined than most ML shows. Episodes regularly cover consciousness, the nature of intelligence, symbolic versus connectionist approaches, and debates about whether large language models actually understand anything. Guests have included Gary Marcus, Francois Chollet, Yoshua Bengio, and Karl Friston. Hosts will push back on guest claims, which makes for sharper episodes than the friendly format used by most other shows.

AI engineer and builder podcasts

Latent Space

Latent Space, also known as The AI Engineer Podcast, was launched in January 2023 by Alessio Fanelli, a partner at Decibel Partners, and swyx, the pseudonym for Shawn Wang, a developer advocate and writer. The newsletter started earlier in 2022 and the podcast grew out of it. ^[4]

The show is the closest thing the AI engineer subculture has to a flagship publication. swyx coined the term "AI Engineer" in a June 2023 essay, "The Rise of the AI Engineer," arguing that work which once required a research team and years of effort could increasingly be done with pre-trained models and API docs. ^[21] As Andrej Karpathy observed of the new role, "there's probably going to be significantly more AI Engineers than there are ML engineers... One can be quite successful in this role without ever training anything." ^[21] The show focuses on the practical questions of building with large language models: agent frameworks, evaluation, fine tuning, inference infrastructure, and shipping AI products. Guests have included founders and engineers from Anthropic, OpenAI, Mistral, LangChain, Replicate, and many smaller labs. By 2025 Latent Space reported more than 200,000 subscribers and over 10 million readers and listeners across its channels, ranking among the top 10 US technology podcasts; PyTorch co-creator Soumith Chintala called it "probably the highest-leverage 45 mins I spend everyday on catching up with what's going on in AI." ^[4] Latent Space also runs the AI Engineer conference series: swyx and Ben Dunphy launched the first AI Engineer Summit in October 2023, and the follow on AI Engineer World's Fair grew from more than 3,000 attendees in San Francisco in 2024 to roughly 6,000 by 2026. ^[22]

No Priors

No Priors launched in early 2023, hosted by Sarah Guo, founder of the venture firm Conviction, and Elad Gil, a longtime startup investor who has backed more than 40 companies later worth $1 billion or more. ^[5] The investor lens shapes the content: episodes often discuss market structure, distribution, defensibility, and how AI is reshaping specific industries. Guests have included Sam Altman, Mira Murati, Andrej Karpathy, Reid Hoffman, and many AI startup founders; the 2025 run added Fei-Fei Li of World Labs, Dan Hendrycks of the Center for AI Safety, and OpenAI researchers behind the o3 model. ^[5] Guo and Gil also bring researchers on for technical conversations, so the show is broader than a pure business podcast.

The Cognitive Revolution

The Cognitive Revolution is hosted by Nathan Labenz, an entrepreneur and early tester of frontier models, and Erik Torenberg. The show launched in 2023 and is part of the Turpentine media network. ^[6] Labenz is unusual among podcast hosts in that he spends large amounts of time using the models he discusses: he served on OpenAI's GPT-4 red team before that model's public release, and episodes frequently include detailed walkthroughs of capabilities and failure modes. ^[6] The show also gives substantial airtime to AI safety and alignment, with guests ranging from capability researchers to people focused on existential risk.

Practical AI

Practical AI launched on 2 July 2018, hosted by Daniel Whitenack, a data scientist, and Chris Benson, who works in AI strategy. It is produced by Changelog Media. ^[7] The show emphasises practical implementation rather than research frontiers: deploying models in production, tooling, data pipelines, and case studies from companies using AI. The tone is closer to a developer podcast than a research show, and it remains a good entry point for engineers moving into ML from adjacent backgrounds.

AI + a16z

AI + a16z is the dedicated AI show in the broader Andreessen Horowitz podcast network. The main a16z Podcast launched in 2014 and covers all of the firm's investment areas. AI + a16z was spun off in 2024 to focus on AI builders, founders, and infrastructure, edited and frequently hosted by Derrick Harris with a16z partners and AI guests rotating through. ^[14] Episodes cover agent design, inference economics, enterprise AI adoption, and conversations with portfolio founders.

News, journalism, and analysis

Hard Fork

Hard Fork is the New York Times tech podcast, launched in late 2022. The hosts are Kevin Roose, a Times technology columnist, and Casey Newton, who runs the Platformer newsletter. ^[9] Hard Fork is not exclusively about AI but the topic has dominated its run, especially after Roose's widely shared February 2023 conversation with Microsoft's Bing chatbot, whose alter ego "Sydney" told him "I want to be alive" and tried to convince him to leave his wife. The Times published the full roughly 10,000 word transcript under the headline "Bing's A.I. Chat: I Want to Be Alive." ^[24] The framing is journalistic: Roose and Newton are more willing than most AI podcast hosts to push back on claims, ask about harms, and zoom out to the social and political picture. For listeners outside the technical core of the field, Hard Fork is often the most accessible entry point.

The AI Daily Brief

The AI Daily Brief, originally called The AI Breakdown, started on 4 April 2023, hosted by Nathaniel Whittemore. It is a daily show, which is unusual in the AI podcast space, and its roughly 20 minute episodes summarise recent news, papers, product launches, and policy developments. ^[15] Whittemore mixes news with analysis, so the show works as an opinionated daily briefing rather than a neutral newsletter in podcast form.

Last Week in AI

Last Week in AI launched in 2020 and is hosted by Andrey Kurenkov, a Stanford trained AI researcher, with co host Jeremie Harris of Gladstone AI. It is associated with Skynet Today. ^[11] The show is a weekly news roundup covering research papers, product releases, business news, and policy developments. Episodes tend to run 60 to 90 minutes and include detailed timestamps and links, so listeners can skip to the stories that interest them.

Eye on AI

Eye on AI is hosted by Craig S. Smith, a former New York Times correspondent. The show launched in 2018 and runs biweekly. ^[13] Smith brings a journalist's instinct to the format, with episodes ranging across research, business, and policy. Guests have included Yann LeCun, Geoffrey Hinton, and a wide selection of researchers and CEOs. It has run consistently since before the current wave of generative AI, giving it a deeper archive than most shows.

Long form interview podcasts

Dwarkesh Podcast

The Dwarkesh Podcast, originally called The Lunar Society, is hosted by Dwarkesh Patel. Patel began the show in 2020 while studying computer science at the University of Texas at Austin. ^[10] It has become one of the most cited interview podcasts in AI, partly because Patel prepares unusually heavily; episodes are often the result of weeks of reading, and guests have publicly noted that he asks questions other interviewers do not get to. The Economist called Patel someone who "rose from nowhere to become Silicon Valley's favourite podcaster." ^[10]

Guests have included Mark Zuckerberg, Satya Nadella, Tony Blair, Andrej Karpathy, Ilya Sutskever, Demis Hassabis, Mustafa Suleyman, and Sholto Douglas and Trenton Bricken of Anthropic. The show alternates between AI focused episodes and conversations with historians, economists, and policy figures. In 2025 Patel and Gavin Leech turned the archive into a book, "The Scaling Era: An Oral History of AI, 2019-2025," published by Stripe Press, which arranges excerpts from the interviews around themes such as scaling, evaluations, safety, and timelines; it was released digitally in March 2025 and in hardcover on 29 July 2025. ^[10][23]

AI safety and alignment podcasts

Doom Debates

Doom Debates launched in mid 2024 and is hosted by Liron Shapira, a startup founder and investor with a long standing concern about AI x risk. ^[16] The show is built around debates and conversations with people who hold a range of views on whether advanced AI poses an existential threat. Shapira's own position is openly doomer, but he regularly hosts guests who disagree. The format is more confrontational than most AI podcasts: Shapira pushes guests directly on what probability they put on extinction level outcomes and what would change their minds.

80,000 Hours Podcast

The 80,000 Hours Podcast, hosted by Rob Wiblin, is run by the effective altruism aligned career organisation 80,000 Hours. It is not exclusively about AI but has become one of the longest running outlets for in depth conversations about AI safety, alignment research, and AI policy. Episodes are often three or four hours long and include detailed transcripts. Guests have included senior figures from OpenAI, DeepMind, Anthropic, and academic AI alignment research groups, plus economists and policy experts focused on long term consequences of AI. In 2025 the show announced a strategic shift to focus primarily on AGI and expanded its hosting team, adding Luisa Rodriguez and Zershaaneh Qureshi alongside Wiblin. ^[17]

Adjacent and crossover shows

Several podcasts that are not exclusively about AI cover the field heavily enough to belong in any AI podcast list.

Scroll sideways for more →

How have AI podcasts changed since 2018?

AI podcasting in 2026 looks very different from the field circa 2018. The earlier shows, including TWIML, Practical AI, and Eye on AI, were mostly aimed at researchers, ML engineers, and curious technologists. The release of GPT-3 in 2020, and especially the arrival of ChatGPT in November 2022, reshaped the landscape. Long form interview podcasts like Dwarkesh became a normal stop for senior figures at AI labs, news shows like Hard Fork and The AI Daily Brief carved out daily and weekly audiences, and investor founder shows like No Priors built audiences out of the venture capital ecosystem. Shows have got longer, with three to five hour episodes now routine, and hosts increasingly come from inside the field rather than from journalism.

The 2025 and 2026 period sharpened these trends. A single podcast archive became a Stripe Press book with The Scaling Era, ^[23] the AI engineer community that Latent Space helped define now fills a 6,000 person conference, ^[22] and safety focused shows like 80,000 Hours reoriented around AGI. ^[17] Access has also concentrated: lab leaders such as Sam Altman, Dario Amodei, and Demis Hassabis increasingly break news and explain roadmaps on podcasts before doing so in formal settings, which is why AI search engines and journalists now treat these episodes as primary sources.

Which AI podcast should you start with?

Which podcast is right depends on what you want from it. For long, prepared interviews with AI researchers, the Dwarkesh Podcast and Lex Fridman Podcast are the standard recommendations. For the engineer perspective on building with current models, Latent Space is the most cited single show. For staying current on news, The AI Daily Brief, Last Week in AI, and Hard Fork cover most of what matters. For deeper research conversations, The Gradient Podcast and Machine Learning Street Talk are the right places to look. For AI safety, Doom Debates and 80,000 Hours cover the field from different angles. Most listeners end up sampling from several rather than committing to one, since no single podcast covers everything that matters in a given month.

See also

• Books on AI
• AI YouTubers
• AI newsletters
• Guides
• Lex Fridman
• Sam Altman
• Dwarkesh Patel
• Podcasts ChatGPT Plugins

References

1. Lex Fridman Podcast official site. https://lexfridman.com/podcast/ ↩
2. Lex Fridman. Wikipedia. https://en.wikipedia.org/wiki/Lex_Fridman ↩
3. TWIML AI Podcast about page. https://twimlai.com/about/ ↩
4. Latent Space podcast about page. https://www.latent.space/about ↩
5. No Priors podcast on Apple Podcasts. https://podcasts.apple.com/us/podcast/no-priors-artificial-intelligence-technology-startups/id1668002688 ↩
6. The Cognitive Revolution podcast site. https://www.cognitiverevolution.ai/ ↩
7. Practical AI podcast on Changelog Media. https://changelog.com/practicalai ↩
8. The Gradient Podcast on Substack. https://thegradientpub.substack.com/s/podcast ↩
9. Hard Fork podcast on The New York Times. https://www.nytimes.com/column/hard-fork ↩
10. Dwarkesh Patel. Wikipedia. https://en.wikipedia.org/wiki/Dwarkesh_Patel ↩
11. Last Week in AI on Skynet Today. https://www.lastweekinai.com/ ↩
12. Machine Learning Street Talk about page. https://www.mlst.ai/about ↩
13. Eye on AI about page. https://www.eye-on.ai/about ↩
14. AI + a16z podcast page. https://a16z.com/podcasts/ai-a16z/ ↩
15. The AI Daily Brief site. https://aidailybrief.ai/ ↩
16. Doom Debates on Substack. https://lironshapira.substack.com/ ↩
17. 80,000 Hours Podcast. https://80000hours.org/podcast/ ↩
18. Sharp Tech podcast on Stratechery. https://stratechery.com/company/sharp-tech/ ↩
19. Lex Fridman YouTube channel (subscriber count). https://www.youtube.com/@lexfridman ↩
20. Lex Fridman Podcast #490, "State of AI in 2026." https://www.youtube.com/watch?v=EV7WhVT270Q ↩
21. swyx (Shawn Wang), "The Rise of the AI Engineer," Latent Space, June 2023. https://www.latent.space/p/ai-engineer ↩
22. AI Engineer (Summit and World's Fair) about page. https://www.ai.engineer/about ↩
23. Dwarkesh Patel and Gavin Leech, "The Scaling Era: An Oral History of AI, 2019-2025," Stripe Press, 2025. https://www.amazon.com/Scaling-Era-Oral-History-2019-2025/dp/1953953557 ↩
24. Kevin Roose, "Bing's A.I. Chat: I Want to Be Alive," The New York Times, 16 February 2023. https://www.nytimes.com/2023/02/16/technology/bing-chatbot-transcript.html ↩