Keeping up with AI genuinely requires active effort. The field moves fast, the terminology shifts constantly, and the gap between hype and reality is wide enough that reading the wrong things can actually leave you more confused than you started. Podcasts, when they’re good, solve a different problem: they let you absorb context while doing something else, and the best ones will change how you think, not just what you know.
This is not an exhaustive list. It’s a curated one. These are the shows I actually listen to, recommend to clients, and find myself returning to when something interesting happens in the field.
The best AI podcasts don’t just tell you what happened. They change how you think about what’s coming.
For Understanding the Technology
Lex Fridman Podcast
Three to four hours long. Worth every minute when the guest is right.
Long-form conversations with researchers, engineers, and founders working at the frontier of AI. Lex talks to people like Sam Altman, Demis Hassabis, Andrej Karpathy, and Yann LeCun in depth — and the conversations often run three or four hours, which means you actually get to the substance rather than the rehearsed talking points. Not every episode is about AI, but when it is, there’s nothing quite like it for genuine depth. The pace suits a commute or a long walk rather than background listening.
The TWIML AI Podcast (This Week in Machine Learning)
Running since 2016 — Sam Charrington has seen every AI hype cycle and calls each one accurately.
Sam Charrington has been running this show since 2016, which in AI terms means he has seen several complete cycles of hype and correction. The guests are technical — researchers, ML engineers, academics — and the conversations go into real depth on how models actually work. If you want to understand transformers, diffusion models, or reinforcement learning from human feedback without a computer science degree, this is where to start. Episodes are typically 45–90 minutes and consistently high quality.
Eye on AI
Research-landscape conversations at 30 to 45 minutes. Long enough to learn, short enough to finish.
Craig Smith interviews AI researchers and practitioners in a format that’s more accessible than TWIML but still substantive. Good for understanding the research landscape without needing a deep technical background. Episodes are shorter (30–45 minutes) and work well as a first introduction to specific topics before going deeper elsewhere.
For Business Strategy and Practical Application
The AI Breakdown
Near-daily, 15 to 30 minutes. The analysis is consistently sharper than most AI newsletters.
Nathaniel Whittemore produces near-daily episodes covering AI news, product releases, and strategic implications. The format is compact — usually 15–30 minutes — and it functions more as an intelligent briefing than a deep dive. If you want to stay current without spending hours on it, this is the most efficient way to do it. The analysis is consistently sharper than most AI newsletters.
No Priors
Founders and executives who have actually deployed AI — telling you what is and is not working.
Co-hosted by Sarah Guo and Elad Gil, both with deep venture capital and operator backgrounds. Guests are typically founders and executives building AI-native companies, and the conversations focus on the business reality of deploying AI: what’s working, what isn’t, what the actual unit economics look like. Less hype, more honest assessment of where AI creates durable value and where it doesn’t. Highly recommended if you’re thinking about AI from a product or investment angle.
Lenny’s Podcast
Operators, not theorists. The AI episodes are some of the most practically useful content available.
Not exclusively about AI, but Lenny Rachitsky has been covering AI’s impact on product development and growth with increasing depth. The practical episodes on how product teams are actually using AI tools day-to-day are some of the most useful content available for anyone thinking about how this changes the way businesses operate. His guests are operators, not theorists, which makes the advice actionable.
For the Bigger Picture
Hard Fork (New York Times)
Honest about uncertainty, willing to say when something is overhyped. Send this to your board.
Kevin Roose and Casey Newton cover tech broadly but AI is a recurring and increasingly central topic. The strength of this show is tone: it’s honest about uncertainty, willing to say when something is overhyped, and good at explaining complex developments to a general audience without dumbing them down. If you have senior leadership or board members who want to understand AI but aren’t going to listen to a technical podcast, this is what to send them.
The AI Podcast (NVIDIA)
Strong guest list, applied focus across industries. Treat the source accordingly.
Yes, it’s produced by NVIDIA, so treat it accordingly. But the guest list is strong and the conversations are genuinely informative about applied AI across industries — healthcare, climate, manufacturing, robotics. Good for understanding where AI is being deployed in the physical world rather than just in software.
Futures of Learning
More thoughtful on AI and workforce development than most shows ten times its size.
A smaller show, but worth including for anyone thinking about AI and workforce development. The conversations about how AI is changing what skills matter, what education needs to look like, and how organisations build internal AI capability are more thoughtful than most. Particularly relevant if you are running AI training programmes for your team.
A Few Honourable Mentions
The Gradient Podcast — Academic and research-focused, excellent for understanding where the field is going rather than where it is. Dense but rewarding.
Practical AI — Friendly, accessible, and genuinely practical. Good starting point if you are new to the topic and want something that assumes no prior knowledge.
Acquired — Not an AI show, but their deep dives on Nvidia, Google, and Microsoft provide essential context for understanding who controls the infrastructure that runs the AI world. The Nvidia episode in particular is essential listening.
Pick two or three shows, listen consistently for a month, and take one idea from each episode that applies directly to your business. That is worth more than trying to follow everything.
What to Actually Do With This List
If you are new to AI and want a starting point: Hard Fork for context, The AI Breakdown for staying current, and No Priors for business application. That combination covers the landscape without overwhelming you.
If you are already technical and want to go deeper: TWIML and Lex Fridman for the research layer, No Priors for the commercial layer.
If you are a business owner trying to work out what AI means for your specific situation: Lenny’s Podcast and No Priors will give you the most direct return. And if you want a structured view of where AI actually applies to your business rather than a general education — that’s what an AI Audit is for.