Best AI podcasts in 2026

Akshay Datt

host of Founder Thesis, a long-form interview podcast with company founders. He writes about how companies use podcasts as a distribution channel.

The best AI podcasts in 2026 are Dwarkesh Podcast for research depth, Latent Space for AI engineering, No Priors for the investor view, The TWIML AI Podcast for applied machine learning, Hard Fork for the weekly news cycle, and The Cognitive Revolution for builder interviews.

An AI podcast is a recurring audio or video show that covers artificial intelligence research, products and business through interviews, news analysis or technical explainers. AI podcasts split into four groups: research interviews, engineering shows, investor and founder shows, and daily news briefings. Which one suits you depends on whether you build with models, invest in them, or sell to people who do.

Top AI podcasts

Best AI podcasts for technical depth

Dwarkesh Podcast and Latent Space carry the most technical depth of any AI podcast. Dwarkesh Patel runs interviews of 75 to 130 minutes with researchers, and guests include Ryan Greenblatt of Redwood Research on automating AI research and Adam Brown on general relativity. The feed also carries short essay episodes of nine to twenty minutes.

Latent Space, hosted by Shawn Wang and Alessio Fanelli, is built for engineers shipping AI systems. Episodes cover inference engineering with Baseten, an AI infrastructure company, Codex adoption at OpenAI, and model training at Poolside. It assumes you already build with models.

Machine Learning Street Talk, run by Tim Scarfe, sits further toward theory and philosophy of mind. The TWIML AI Podcast, hosted by Sam Charrington, has the largest technical back catalogue, though it now publishes one episode every two to three weeks.

Practical AI, hosted by Daniel Whitenack and Chris Benson, covers implementation questions closer to the data pipeline than the research frontier.

Best AI podcasts for founders and investors

No Priors, Training Data, Unsupervised Learning and AI + a16z are the four AI podcasts made for founders and investors. All four are produced by venture firms and interview operators rather than academics.

No Priors is hosted by Sarah Guo, founder of the venture firm Conviction, and investor Elad Gil. Guests include the CEOs of Chess.com and Booking.com and the co-founders of DoorDash. The conversations run on company building and market structure.

Training Data comes from Sequoia Capital and mixes researcher interviews with founder conversations. Unsupervised Learning is hosted by Jacob Effron of Redpoint Ventures and runs closer to product decisions inside AI startups. AI + a16z, from Andreessen Horowitz, leans toward infrastructure.

The Cognitive Revolution serves the same audience. Nathan Labenz, founder of the AI video company Waymark, interviews builders at a high volume.

Best AI podcasts for marketers and business teams

The Artificial Intelligence Show, How I AI and Me, Myself, and AI are the strongest AI podcasts for people applying AI to commercial work rather than building models. None of them require a machine learning background.

The Artificial Intelligence Show is hosted by Paul Roetzer and Mike Kaput of Marketing AI Institute, an education company for marketers. It runs weekly and treats AI as an adoption problem inside companies.

How I AI is hosted by Claire Vo, founder of ChatPRD, an AI writing tool for product specs. Episodes run about thirty minutes and are built around a screen share of a real workflow, which makes the steps directly copyable.

Me, Myself, and AI comes from MIT Sloan Management Review and works through enterprise AI adoption case by case. It publishes less often than the others.

Everyday AI, hosted by Jordan Wilson, publishes daily and is aimed at business users learning tools rather than architecture.

Best AI news podcasts, daily and weekly

The AI Daily Brief, Hard Fork and Last Week in AI cover the AI news cycle at three different speeds. All three together run about five hours a week, so most listeners pick one.

The AI Daily Brief, hosted by Nathaniel Whittemore, publishes every weekday in short episodes. It is the fastest way to know what happened without reading a feed.

Hard Fork is hosted by Kevin Roose and Casey Newton and produced by The New York Times. It runs weekly, mixes news segments with interviews, and covers policy and consumer impact more than implementation.

Last Week in AI, hosted by Andrey Kurenkov and Jeremie Harris, is the densest weekly roundup of model releases, papers and safety developments.

What is the Human By Design podcast?

Human By Design is an AI podcast hosted by Dr. Arjun Jain, founder and chief scientist of Fast Code AI, an AI research and engineering company. It covers AI research and the work of building AI systems, with a focus on how established companies move models into production.

The first episode is Jain explaining how to build a foundation model. Later episodes include Jürgen Schmidhuber, whose work underpins modern sequence models, and Simon Jagers of Samotics, an industrial monitoring company, on why data quality outranks algorithm choice.

Episodes run long, between one and two hours, apart from one short entry.

Disclosure: Human By Design is produced by Unnati, the studio that publishes this blog.

Dr. Jain holds a PhD in computer graphics from the Max Planck Institute for Informatics, did postdoctoral research at NYU’s Courant Institute, and was a research scientist at Apple before founding Fast Code AI.

How to check if a podcast is active

Check the date of the most recent episode in the AI podcast’s RSS feed or on Apple Podcasts. Many shows that ranked well in 2024 stopped releasing episodes without announcing it.

Four checks separate a live show from an abandoned one:

1. Latest episode date. Anything older than 90 days signals a pause or an end.
2. Gap pattern. Compare the last six episode dates. Widening gaps come before a stop.
3. Feed versus platform. Apple and Spotify listings can lag the RSS feed by days.
4. Season structure. Some shows publish in seasons. Google DeepMind: The Podcast, hosted by Hannah Fry, runs this way, so a multi-month gap is normal rather than terminal.

Why do company podcasts fail?

Company podcasts fail on three production choices: a broad audience, a schedule the host cannot hold, and guests the host cannot question hard.

Guest selection does most of the work. Dwarkesh Patel prepares to the point where researchers answer questions they do not get elsewhere, and that preparation, not the audio quality, carries the length.

Cadence separates the durable shows from the abandoned ones. The AI Daily Brief and Everyday AI both publish daily, on a schedule the audience can rely on. A company podcast that stops after a handful of episodes has built nothing.

Narrow beats broad. Latent Space addresses AI engineers and nobody else. A show aimed at everyone in technology competes with Hard Fork and loses.

Frequently asked questions