The Best AI Podcasts for Staying Informed
19 AI and tech podcasts sorted by the job each does, with checked audio feeds, notes on which are still active, and an OPML download.
This is a list of 19 AI podcasts, sorted by the job each one does: explain a technical idea, show how a team built something, or put a news story in context. Start with the five below, sample one episode of each, and keep the ones that answer questions you actually have.
Importing the whole list doesn't mean you have to listen to all of it.
Checked September 17, 2026: each show's identity, its feed file, and the audio dates the publisher supplies. The notes say which shows are still active and which are archives that still work.
Last verified Sep 17, 2026. Recheck target: monthly. The OPML holds 19 available feeds from 19 listed sources. Entries without a verified feed URL are excluded from the import.
A show makes the list when the host understands the topic, the guests do the work instead of making the podcast rounds, and not much time goes to intros and ads.
Start with five
One long interview show, two builder shows, a news brief and an applied machine learning show.
Long interviews with researchers and founders, with transcripts so you can follow the technical argument. Start with a guest whose work you already want to understand.
Interviews with builders about agents, models and the infrastructure underneath. Good for hearing the engineering questions behind a product announcement.
Nathan Labenz interviews builders and researchers about how AI systems work and where they're being used. A good companion to hands-on, how-to-build-it reading.
Nathaniel Whittemore covers AI news and what it means for business. Use it to spot a story worth digging into, then go read the original announcement or reporting.
Daniel Whitenack and Chris Benson talk about applied AI systems and engineering decisions. Pick an episode tied to a problem you can test in your own project.
The directory
Skip the non-AI episodes of Lex, Lenny, All-In and Acquired. Treat investor roundtables as a mood check, not evidence.
Long interviews with researchers and founders, with transcripts so you can follow the technical argument. Start with a guest whose work you already want to understand.
The official site advertises this mixed feed: 15 of its 20 checked items have audio enclosures. Text-only entries may appear differently across podcast clients. Activity uses the latest audio item.
Interviews with builders about agents, models and the infrastructure underneath. Good for hearing the engineering questions behind a product announcement.
The official podcast page advertises this dedicated audio feed. All 223 checked entries have audio enclosures; the 13.6 MB feed required a larger audit limit. Activity is publisher supplied.
Sarah Guo and Elad Gil talk with founders and researchers about AI companies, products and how they reach customers. Get the business side here, then read the technical sources too.
The public feed parsed successfully; its title and publisher match this show. The activity date is the latest audio item timestamp supplied by the publisher.
Pick the AI and programming episodes from a much broader show. The long format leaves room for technical detail, but the full feed is not an AI news brief.
The public feed parsed successfully; its title and publisher match this show. The activity date is the latest audio item timestamp supplied by the publisher.
Nathaniel Whittemore covers AI news and what it means for business. Use it to spot a story worth digging into, then go read the original announcement or reporting.
The public feed parsed successfully; its title and publisher match this show. The activity date is the latest audio item timestamp supplied by the publisher.
Michael and Chris Sharkey talk through AI releases and hands-on experiments in a relaxed way. The checked feed has had no new episode since May 8, 2026, so treat it as an archive.
The feed and Apple catalog identify Michael and Chris Sharkey. The latest audio timestamp is May 8, 2026; the checked feed does not support a claim of current weekly publishing.
AI segments inside a broader tech show. Good for the business side, not for how systems are built.
The public feed parsed successfully; its title and publisher match this show. The activity date is the latest audio item timestamp supplied by the publisher.
Nathan Labenz interviews builders and researchers about how AI systems work and where they're being used. A good companion to hands-on, how-to-build-it reading.
The public feed parsed successfully; its title and publisher match this show. The activity date is the latest audio item timestamp supplied by the publisher.
Daniel Whitenack and Chris Benson talk about applied AI systems and engineering decisions. Pick an episode tied to a problem you can test in your own project.
The public feed parsed successfully; its title and publisher match this show. The activity date is the latest audio item timestamp supplied by the publisher.
Interviews about how large companies adopt AI, run it and get value from it. Useful for learning what buyers care about, but a guest's results won't necessarily carry over to your project.
The public feed parsed successfully; its title and publisher match this show. The activity date is the latest audio item timestamp supplied by the publisher.
Sam Charrington has been doing technical ML interviews longer than most AI podcasts have existed.
The public feed parsed successfully; its title and publisher match this show. The activity date is the latest audio item timestamp supplied by the publisher.
Craig S. Smith interviews AI researchers, founders and operators. Pick by guest and topic, since the archive mixes technical work with company stories.
The public feed parsed successfully; its title and publisher match this show. The activity date is the latest audio item timestamp supplied by the publisher.
Kevin Roose and Casey Newton cover tech as reporters. Good for the business and culture around an AI release.
The NYT show identity matches the Apple catalog and feed. The public feed exposed only three audio entries when checked, so this import does not provide the full archive.
Product and growth interviews, some with AI product teams. Pick the episodes that fit rather than expecting the whole feed to be about AI engineering.
The public feed parsed successfully; its title and publisher match this show. The activity date is the latest audio item timestamp supplied by the publisher.
AI episodes only, and even then as a read on what investors believe, not as evidence.
The public feed parsed successfully; its title and publisher match this show. The activity date is the latest audio item timestamp supplied by the publisher.
Company histories that explain how products reach customers, how businesses make money and the infrastructure behind them. Pick the tech episodes that fit, since this is not a weekly AI news show.
The public feed parsed successfully; its title and publisher match this show. The activity date is the latest audio item timestamp supplied by the publisher.
Long, argumentative conversations with researchers. Not a commute show unless you want the argument.
The public feed parsed successfully; its title and publisher match this show. The activity date is the latest audio item timestamp supplied by the publisher.
Andrey Kurenkov and Jeremie Harris go through recent AI news, research and policy. This is the Skynet Today show tied to lastweekin.ai, not the separate AI Weekly show with the same name.
Matched the Skynet Today feed to Apple show 1502782720 and the episode credits for Andrey Kurenkov and Jeremie Harris. Excluded the different same-name AI Weekly show.
AI episodes on careers, catastrophic risk and what to do with a working life. Listen when your question isn't about a model.
The public feed parsed successfully; its title and publisher match this show. The activity date is the latest audio item timestamp supplied by the publisher.
Pair a weekly interview with the newsletter directory so you're not getting the same five headlines in two formats. When a guest mentions a paper, open the blog and news stack.
Import the feeds, then choose what to keep
Download the OPML file above and use your podcast app's OPML import option, if it has one. The file includes a public feed for all 19 shows. Check the subscriptions it creates before you turn on automatic downloads. Apps handle mixed feeds and partial archives differently.
Three details matter:
- Dwarkesh: the official feed mixes audio and text posts. The check found 15 entries with audio among the 20 items it returned.
- Hard Fork: the public feed returned three audio entries. Importing it doesn't get you the full archive.
- This Day in AI: the newest audio item in the checked feed is dated May 8, 2026. Keep it for the archive, not as a current weekly recap.
The date next to each entry comes from the publisher's feed. It shows a dated episode exists. It doesn't mean every episode was listened to or that its claims were checked. Feed identity and file parsing were checked. Audio playback and imports in specific apps weren't tested.
Change log
Expanded OPML from 3 to 19 verified audio-bearing feeds. Updated moved show links, distinguished same-name podcasts, and recorded feed activity and archive limitations. XML parsing and identity checks do not establish playback in every client.
Rechecked exported feeds and show links. Corrected This Day in AI to a weekly cadence and replaced the unavailable No Priors website with its official show links.
Moved onto the directory template with start-with-five, per-entry cadence, and an OPML download of show feeds.
Follow changes to every directory with the directory changes RSS feed.