Table of Contents
Quick Answer
Top 3 free AI podcasts for 2026:
Latent Space — practitioner-focused, weekly
The TWIML AI Podcast — research + industry interviews
Dwarkesh Patel — long-form interviews with AI leaders
All podcasts are free; optional premium tiers exist
Listed by signal-to-noise, not popularity
Most have text transcripts
Why These Resources Matter
Podcasts are the best way to absorb AI context during commutes and walks. The interviews below frequently break news before mainstream coverage and include deep technical discussion.
The List
Latent Space (latent.space) — Swyx + Alessio. Practitioner-heavy.
The TWIML AI Podcast (twimlai.com) — Sam Charrington. Long-running, high signal.
Dwarkesh Podcast (dwarkeshpatel.com) — Long, research-heavy interviews.
Practical AI (changelog.com/practicalai) — Engineering focus.
Lex Fridman Podcast (lexfridman.com) — Broad, sometimes AI-focused.
The Gradient Podcast (thegradient.pub) — Academic.
Machine Learning Street Talk (youtube.com + podcast) — Tim Scarfe.
No Priors (no-priors.com) — Sarah Guo + Elad Gil.
Eye on AI (eye-on.ai) — Craig Smith interviewing notable researchers.
The Cognitive Revolution (cognitiverevolution.ai) — Nathan Labenz.
AI Daily Brief (NLW) (nlwpodcast.com) — 10-minute daily updates.
Gradient Dissent (wandb.ai/site/podcast) — Weights & Biases.
Interconnects (interconnects.ai) — Nathan Lambert, RLHF focus.
NeurIPS Podcasts — Post-conference interviews, free.
AI Alignment Podcast (futureoflife.org) — Safety and policy.
Import AI (audio) (jack-clark.net) — Jack Clark summary.
DeepMind Podcast (deepmind.google/podcast) — Official DeepMind.
NVIDIA AI Podcast (blogs.nvidia.com/ai-podcast) — Industry stories.
How to Get the Most Out of These Resources
- Subscribe to 3–4 shows; avoid podcast overload
- Use 1.25x–1.5x playback for most interviews
- Read show notes after; they link to papers
- Pick episodes on topics you already half-understand for best retention
Next Steps / Advanced Resources
When you want original sources: read arXiv papers referenced in the episode; most shows link them.
FAQs
Best for news? NLW's AI Daily Brief.
Best for technical depth? Latent Space + TWIML.
Best for career? No Priors + Gradient Dissent.
Best for beginners? Practical AI + Eye on AI.
Do they have transcripts? Most, yes — helpful for citation.
How long per episode? 30–180 minutes.
Conclusion
Pick three podcasts, listen during walks for two weeks, and drop any that bore you. Signal over loyalty.