Hundreds of AI podcasts launched in the past two years. I've sampled more of them than I'd like to admit. Most are recycled news headlines with mediocre commentary — someone reading an OpenAI press release and adding "this is huge" before moving on. You deserve better than that.
These 10 consistently deliver real insight. Whether you want to understand the tech, build an AI business, or think seriously about where all of this is heading, there's something here for you. I've been listening to all of them for at least six months, most for over a year. They've earned the recommendation.
For Understanding AI Technology
1. Lex Fridman Podcast
Host: Lex Fridman (MIT researcher)
Episode length: 2-4 hours
Frequency: 1-2 per week
Lex's format is the opposite of everything modern media rewards: long, slow, unstructured conversations where guests actually finish their thoughts. When Sam Altman sits down for three hours, you hear things he'd never say in a 5-minute CNBC hit. When Dario Amodei explains Anthropic's safety approach, you get the full reasoning, not a soundbite.
The episodes are a commitment. Three hours is a lot. But here's what I've found: the best insights always come after the 90-minute mark, once the guest has exhausted their talking points and starts saying what they actually think. That's when you hear the off-script admissions — the stuff that matters.
Not for your commute unless you like multi-part listening. For a long drive, a workout day, or a weekend deep-dive, there's nothing better.
2. The AI Podcast (NVIDIA)
Host: Noah Kravitz
Episode length: 20-40 minutes
Frequency: Weekly
This one flies under the radar because it's produced by NVIDIA and people assume it's corporate fluff. It's not. The focus is practical: how real companies use AI in production. Not theoretical possibilities — actual deployed systems, actual results, actual problems encountered along the way.
The episode lengths are manageable and the signal-to-noise ratio is excellent. If you want to understand what AI looks like when it leaves the lab and enters a warehouse, a hospital, or a trading floor, this is where you go.
3. Practical AI (Changelog)
Host: Chris Benson and Daniel Whitenack
Episode length: 45-60 minutes
Frequency: Weekly
Bridges the gap between research papers and production deployment. If you're an engineer or developer building AI systems — or if you manage people who are — this is your podcast. Technical without being exclusionary. They explain concepts clearly without dumbing them down to the point of uselessness, which is a harder balance to strike than most hosts manage.
Their episodes on MLOps and model deployment have saved me hours of googling. Genuinely useful if you're in the trenches.
For AI Business & Strategy
4. Hard Fork (NYT)
Host: Kevin Roose and Casey Newton
Episode length: 45-60 minutes
Frequency: Weekly
The most fun AI-adjacent podcast. Period. Kevin Roose and Casey Newton genuinely enjoy arguing with each other, and their chemistry makes complex topics feel like overhearing a smart dinner conversation. You actually look forward to new episodes, which is not something I say about many shows covering enterprise AI strategy.
What makes it valuable beyond entertainment: they contextualize every AI development within the broader tech business landscape. When a new model drops, they don't just explain what it does — they explain who wins, who loses, and what it means for the companies you care about. Their coverage of the AI workforce impact has been particularly sharp.
If someone told me I could only subscribe to one tech podcast, this would be it.
5. AI Breakdown
Host: Nathaniel Whittemore
Episode length: 15-30 minutes
Frequency: Daily
The only daily AI podcast worth subscribing to. Nathaniel distills the day's important AI developments into 15-20 minutes of genuinely useful commentary. No filler. No "well, it remains to be seen" hedging. He picks a stance and defends it.
I listen to this on my morning routine. By the time I sit down at my desk, I know what happened in AI yesterday and have a framework for thinking about it. That's worth more than any newsletter.
The daily format means some episodes are better than others — inevitable when you produce content every single day. But his hit rate is remarkably high. Seven out of ten episodes teach me something I didn't know.
6. The Cognitive Revolution
Host: Nathan Labenz
Episode length: 60-90 minutes
Frequency: 2 per week
Here's what separates Nathan from most AI podcast hosts: he actually built AI products. He's not commentating from the sidelines. He's been in the weeds, fought with production systems, dealt with the gap between demo and deployment. His questions hit different because they come from experience.
The guest selection is unusually good — less "famous name doing a press tour" and more "person who actually built the thing and can explain what went wrong." If you're building an AI startup or managing AI integration at a larger company, the tactical insights here are unmatched.
For AI and Society
7. Eye on AI
Host: Craig S. Smith
Episode length: 30-45 minutes
Frequency: Weekly
The most consistently thoughtful podcast about AI's broader implications. Craig doesn't chase hype cycles. He invites researchers, ethicists, and policymakers for measured conversations about what AI means for society, governance, and economics.
I started listening because I was tired of every AI podcast being either "AI will save the world" or "AI will destroy the world." Eye on AI occupies the nuanced middle ground that reality actually lives in. If you want to sound informed rather than ideological when AI comes up in conversation, this is your source.
8. Your Undivided Attention (Center for Humane Technology)
Host: Tristan Harris and Aza Raskin
Episode length: 45-60 minutes
Frequency: Biweekly
Essential counterbalance to the AI hype machine. Tristan Harris and Aza Raskin ask the questions that AI companies would prefer you not think about. How are incentive structures shaping AI development? Who bears the cost when things go wrong? What happens to democracy when synthetic media is cheap and verification is expensive?
Some people find this podcast alarmist. I find it bracing. Every industry needs critics who push back on the prevailing narrative, and these two do it with more rigor and less hysteria than anyone else I've found. Even when I disagree with their conclusions, I'm grateful someone is asking the questions.
For AI Makers and Builders
9. Latent Space
Host: Alessio Fanelli and Swyx
Episode length: 60-90 minutes
Frequency: Weekly
The most technically rigorous AI podcast. If you're building AI products or infrastructure, subscribe and listen weekly. Full stop.
What sets Latent Space apart: the hosts have enough technical depth to push back on guests and ask the second-level questions. They don't just ask "what does your product do?" — they ask "why did you choose that architecture? What broke when you scaled? What would you do differently?" That level of honesty about technical tradeoffs is rare and incredibly valuable if you're making similar decisions.
Their episodes on AI agent architectures and LLM inference optimization are the best I've found anywhere, including paid courses.
10. Last Week in AI
Host: Andrey Kurenkov
Episode length: 45-60 minutes
Frequency: Weekly
If you only have time for one AI podcast per week, this covers everything important. Andrey does a comprehensive roundup of the week's AI developments — research, industry, policy, culture — with enough context that you don't need to have followed the daily news.
Think of it as your weekly briefing. I listen to this on Sunday evenings and start Monday feeling caught up. The format is efficient, the analysis is balanced, and Andrey doesn't waste your time on developments that don't matter. That editorial judgment — knowing what to leave out — is what makes it worth the subscription.
How to Pick
| If you want... | Listen to... |
|---|---|
| Deep understanding | Lex Fridman, Latent Space |
| Daily news | AI Breakdown |
| Weekly roundup | Hard Fork, Last Week in AI |
| Business strategy | The Cognitive Revolution |
| Ethics and society | Your Undivided Attention, Eye on AI |
| Technical depth | Practical AI, Latent Space |
My suggestion: start with Hard Fork (most accessible) or AI Breakdown (shortest commitment). Add others based on your interests. Three to four active subscriptions is the sweet spot — more than that becomes noise rather than signal.
---
Keep Reading
- 12 AI YouTube Channels Worth Your Time — when you want to watch instead of listen
- 7 AI Books That Change How You Think — when you want to go deeper than any podcast can take you
- Cursor AI Freelancer Guide — put what you learn from these pods into freelance income
- ChatGPT vs Claude vs Gemini — the tools every podcast is talking about, compared head-to-head
