Dwarkesh Podcast

by Dwarkesh Patel · Dwarkesh Patel

Long-form interviews with AI researchers, lab leaders and other specialists, hosted by Dwarkesh Patel. Irregular, typically one to several episodes a month, with occasional essay readings. Suits listeners wanting technical depth on AI progress and alignment.

podcastinterviewstechnicalresearchfree
Read at Sourcewww.dwarkesh.com

At a Glance

Format

Podcast

Author

Dwarkesh Patel

Source

Dwarkesh Patel

Highlights

What it is

The Dwarkesh Podcast is a long-form interview show hosted by Dwarkesh Patel. Its Apple Podcasts description calls it "deeply researched interviews". Recent episodes centre on AI: agent swarms, recursive self-improvement, AI compute economics and alignment risk. It also strays into physics and maths.

What an episode looks like

Most episodes are a single guest for roughly one to two and a half hours. Recent guests include Noam Brown (17 September 2026), Ajeya Cotra (1 September), Dylan Patel (25 August) and Ryan Greenblatt (11 August). There are also panel discussions, such as the 11 September episode with several AI researchers, and occasional narrated or video versions of the host's written essays, such as "The rise and fall of agent civilizations" (31 August).

Cadence and back catalogue

Apple Podcasts UK showed 141 episodes on 30 September 2026. The feed's ten most recent items ran from 30 June to 17 September 2026, so output is irregular: a gap of about three and a half weeks separated 10 July from 3 August, and there were two episodes in some weeks (31 August and 1 September). The feed is public and free to subscribe to in any podcast app.

Who it suits

Listeners who already follow AI and want guests questioned at length on technical detail, rather than news summaries. Expect heavy use of jargon. Episodes are also published on the show's Substack, which the feed says offers bonus episodes and subscriber discussion. The price of the paid tier is not shown.

Last updated on