ChatGPT 
Summary
Key takeaways
- The article curates more than 40 AI podcasts and focuses on shows that remained active in 2026 rather than recycling outdated recommendations.
- The strongest podcast choices depend on role: engineers, CTOs, researchers, founders, data teams, and policy-focused leaders need different listening stacks.
- Latent Space is positioned as one of the strongest choices for AI engineers working on agents, infrastructure, multimodality, and production systems.
- Dwarkesh Podcast and Machine Learning Street Talk are better suited to listeners who want deep technical discussion, frontier-lab interviews, and research-level context.
- The AI Daily Brief is the strongest short-format option for keeping up with fast-moving AI news without spending hours on long-form episodes.
- The Cognitive Revolution, No Priors, and similar shows are especially useful for CTOs, founders, and executives translating technical AI developments into business and organizational decisions.
- For production AI, podcasts focused on MLOps, observability, evaluation, data infrastructure, and real deployment experience are usually more valuable than generic AI commentary.
- Python-focused shows such as Talk Python To Me and Python Bytes are increasingly relevant because Python remains central to applied AI engineering.
- A small, role-specific listening stack is more useful than subscribing to dozens of feeds and accumulating a backlog that never gets consumed.
- The article recommends rotating podcasts over time as technologies, hosts, and editorial focus change rather than treating any list as permanently fixed.
When this applies
This applies when an engineer, CTO, founder, researcher, or technical leader wants a curated way to keep up with AI without relying only on news articles, social media, or vendor content. It is especially useful when the goal is to follow frontier research, production AI engineering, agents, MLOps, AI strategy, policy, Python, or data engineering through long-form conversations with practitioners and researchers. It also works well for people who want to build a focused weekly listening routine around their specific role.
When this does not apply
This does not apply as directly when the goal is to learn AI systematically from first principles, complete a structured technical course, or find implementation-level documentation for a specific framework or API. Podcasts are better for context, expert reasoning, trends, and practical experience than for step-by-step technical instruction. They are also less useful when a listener needs immediate, highly specific answers that would be better served by documentation, papers, or hands-on experimentation.
Checklist
- Define whether your main goal is engineering, research, leadership, news, policy, or data.
- Choose three to five podcasts rather than subscribing to every recommended show.
- Include at least one show focused on production AI engineering.
- Add one research-oriented podcast if understanding frontier models and papers matters to your role.
- Add one short-format news podcast if you need frequent updates.
- Choose a strategy or leadership show if you make product or organizational decisions.
- Include an MLOps or infrastructure podcast if you operate models in production.
- Add Python-focused content if your engineering stack is Python-heavy.
- Prefer shows that publish consistently and remain active.
- Check whether recent episodes still match the show’s original focus.
- Prioritize hosts who regularly interview practitioners, researchers, and engineering leaders.
- Use episode length and publishing frequency to build a realistic listening routine.
- Remove shows that remain unplayed for several weeks instead of keeping an unlimited backlog.
- Rotate your listening stack every quarter as your role and priorities change.
- Use podcasts for context and perspective, then verify important technical claims through primary sources.
Common pitfalls
- Subscribing to too many podcasts and creating a backlog that becomes impossible to manage.
- Choosing shows by popularity rather than relevance to your role.
- Relying on generic AI business commentary when you need production engineering depth.
- Using podcasts as a replacement for documentation, research papers, or hands-on technical work.
- Continuing to follow dormant or low-signal shows out of habit.
- Expecting one podcast to cover engineering, research, business, policy, and news equally well.
- Ignoring episode length and frequency when building a listening routine.
- Taking speculative opinions or predictions as established technical facts.
- Listening only to vendor-produced shows without balancing them with independent sources.
- Failing to update your podcast stack as the AI ecosystem and your own responsibilities evolve.
Quick answer: The best AI podcasts in 2026 are Latent Space for AI engineers, Dwarkesh Podcast for long interviews with frontier-lab leaders, Machine Learning Street Talk for technical depth, The AI Daily Brief for daily news and The Cognitive Revolution for CTOs and business leaders. For beginners, start with Hard Fork or Everyday AI. We checked every show on this list for new episodes in September 2026. This shortlist is the working list that the engineering team at Uvik Software uses internally and shares with clients.
Key takeaways
- The number one AI podcast depends on your role. For engineers who ship AI, it is Latent Space. For a general audience, it is Hard Fork.
- The top 3 AI podcasts for most technical listeners in 2026 are Latent Space, Dwarkesh Podcast and Machine Learning Street Talk.
- Beginners and business listeners get more from short, plain-language shows: Everyday AI, The Artificial Intelligence Show and Mixture of Experts.
- Keep three to five shows, not twenty. Add one daily news show for speed.
- Every show on this list published at least one new episode in the 90 days before our September 2026 check.
Top 10 AI podcasts in 2026: the essential shortlist
If you start from zero, subscribe to these ten. The rest of the article expands this shortlist by role and interest.
| # | Podcast | Host(s) | Best for | Frequency |
|---|---|---|---|---|
| 1 | Latent Space | swyx, Alessio Fanelli | AI engineers shipping to production | Weekly |
| 2 | Dwarkesh Podcast | Dwarkesh Patel | Long interviews with frontier-lab leaders | Weekly |
| 3 | Machine Learning Street Talk | Tim Scarfe, Keith Duggar | Technical depth on papers and mechanisms | Fortnightly |
| 4 | The AI Daily Brief | Nathaniel Whittemore (NLW) | Daily AI news at commute length | Daily |
| 5 | The TWIML AI Podcast | Sam Charrington | Research interviews since 2016 | Weekly |
| 6 | Practical AI | Chris Benson, Daniel Whitenack | Applied AI, MLOps, real deployments | Weekly |
| 7 | The Cognitive Revolution | Nathan Labenz | CTOs translating AI to the business | Weekly |
| 8 | No Priors | Sarah Guo, Elad Gil | AI founder and investor lens | Weekly |
| 9 | Interconnects | Nathan Lambert | Open models, RLHF, post-training | Weekly |
| 10 | The Pragmatic Engineer | Gergely Orosz | AI and software engineering at Big Tech | Weekly |
Building with AI? Add senior engineers who ship it.
Uvik Software places senior Python and AI engineers in your team. Matched profiles within 48 hours after a signed SOW.
How we checked: every podcast on this list published at least one new episode in the 90 days before our September 2026 check. We removed dormant shows, shows that moved away from AI and hype-heavy shows. What remains is a list we recommend to our own engineers.
If you lead an engineering team, build AI systems in Python or work in a CTO-level role, this is the working list the engineering team at Uvik Software uses and shares with clients.
Why AI podcasts are different in 2026
Audio is where frontier researchers, founders and operators speak most openly, often months before the same ideas reach papers or the press. The pace of AI is the reason. METR measured the length of software tasks that AI agents can complete, and found that it doubled about every seven months between 2019 and early 2025. Written media cannot keep up with that pace. For engineering leaders, podcasts are one of the fastest ways to hear primary-source reasoning from the people who build these systems.
The list below is organized around depth, freshness and role fit. The useful question is not which show is most popular. It is whether a show helps you understand a paper, ship an agent, improve production reliability, govern a deployment or translate frontier research into a business decision.
Technical and research-grade AI
Deep, long-form, technical listening for people who want to understand papers, mechanisms, research debates and the direction of frontier AI.
Machine Learning Street Talk (MLST)
Hosted by Tim Scarfe and Keith Duggar. A technically focused show for listeners who want long discussions of AI papers, mechanisms, cognitive science, neuroscience, and related research debates. New episodes arrive roughly fortnightly.
Listen: www.mlst.ai
Dwarkesh Podcast
Hosted by Dwarkesh Patel. Long, carefully researched interviews with frontier AI researchers, lab leaders, and other technical guests. It is a strong choice when you want primary-source thinking rather than a short news recap.
Listen: www.dwarkesh.com/podcast
Lex Fridman Podcast
Hosted by Lex Fridman. The show is broader than AI, but its AI episodes regularly run for several hours and explore models, scaling, coding, agents, compute, and the direction of frontier research.
Listen: lexfridman.com/podcast
The TWIML AI Podcast
Hosted by Sam Charrington. A long-running interview show for machine learning researchers, data scientists, and engineering leaders, with recurring coverage of research, production systems, and the changing LLM landscape.
Listen: twimlai.com
Interconnects
Hosted by Nathan Lambert. The feed focuses on open models, RLHF, post-training, and the wider frontier AI ecosystem, combining interviews with audio versions of research-driven analysis.
Listen: www.interconnects.ai/podcast
Google DeepMind: The Podcast
Hosted by Hannah Fry. A behind-the-scenes look at DeepMind research, including areas such as AlphaFold, robotics, world models, nature, and cybersecurity.
Listen: deepmind.google/the-podcast
AI engineering and applied AI for builders
For people shipping AI into production: agents, RAG, evals, inference, infrastructure and deployment. If you write code that calls models or operate the systems around them, start here.
Latent Space: The AI Engineer Podcast
Hosted by swyx (Shawn Wang) and Alessio Fanelli. A technical show for engineers building with foundation models, agents, infrastructure, code generation, and multimodal systems. It is especially useful for people shipping AI into production.
Listen: www.latent.space/podcast
Practical AI
Hosted by Chris Benson and Daniel Whitenack. The show concentrates on applied AI, MLOps, LLMs, edge AI, and deployment questions, with an emphasis on systems that actually reach production.
Listen: practicalai.show
Vanishing Gradients
Hosted by Hugo Bowne-Anderson. Long-form conversations on agents, evaluation, multimodal systems, data infrastructure, and the engineering practices behind reliable AI systems.
Listen: hugobowne.substack.com/podcast
Gradient Dissent
Hosted by Lukas Biewald and produced by Weights & Biases. Conversations with AI leaders focus on the engineering choices and operational realities behind deployed ML and AI systems.
Listen: wandb.ai/site/resources/podcast
MLOps, ML platforms, and production systems
For platform teams, ML infrastructure engineers and anyone responsible for keeping AI reliable in production.
MLOps Community Podcast
Hosted by Demetrios Brinkmann. Practitioner conversations about getting AI and machine learning into production, covering traditional ML, LLMs, agents, infrastructure, tooling, and reliability.
Listen: podcast.mlops.community
AI strategy and enterprise AI for leaders
These shows connect technical capability with organizational and business implications. They are useful for CTOs and engineering leaders who need to translate AI into decisions.
The Cognitive Revolution
Hosted by Nathan Labenz and Erik Torenberg. Interviews with AI builders that connect frontier technical capability with the organizational and business implications leaders need to understand.
Listen: www.cognitiverevolution.ai
Eye on A.I.
Hosted by Craig S. Smith. Interviews that put new AI advances into broader context and help technical leaders distinguish durable shifts from incremental announcements.
Listen: www.eye-on.ai/podcast-archive
Me, Myself, and AI
Hosted by Sam Ransbotham and associated with MIT Sloan Management Review. The show examines how organizations deploy, manage, and govern AI, with an emphasis on what separates real adoption from hype.
Listen: sloanreview.mit.edu/audio-series/me-myself-and-ai
Hard Fork
Hosted by Kevin Roose and Casey Newton. A weekly technology show with substantial AI coverage, useful for leaders who need to track how AI is landing in products, policy, culture, and the public conversation.
Listen: www.nytimes.com/column/hard-fork
Pioneers of AI
Hosted by Rana el Kaliouby. Conversations with creators, critics, researchers, founders, and other people shaping AI, framed around how the technology is changing business and everyday life.
Listen: www.pioneersof.ai
AI founder and investor shows
Useful for following capital, startups, platform shifts and competitive dynamics around AI.
No Priors
Hosted by Sarah Guo and Elad Gil. Interviews with founders, researchers, and leaders across frontier labs, AI-native startups, and large technology companies, viewed through a founder and investor lens.
Listen: www.no-priors.com
Training Data
Hosted by members of the Sequoia Capital team, including Sonya Huang, Pat Grady, and Stephanie Zhan. Conversations with AI builders and researchers about product, research, infrastructure, and company-building decisions.
Listen: www.sequoiacap.com/series/training-data
AI + a16z
Produced by Andreessen Horowitz. Discussions with AI engineers, founders, and investors about how the technology is evolving and how new products and infrastructure are being built around it.
Listen: a16z.com/podcasts/ai-a16z
Unsupervised Learning
A Redpoint Ventures podcast hosted by Jacob Effron alongside Patrick Chase, Jordan Segall, and Erica Brescia. It focuses on what is real in AI now, what may come next, and what the changes mean for builders, researchers, and investors.
Listen: unsupervised-learning.simplecast.com
Possible
Hosted by Reid Hoffman and Aria Finger. An optimistic but practical look at technology, policy, business, and society, with frequent discussion of AI and the choices that shape its impact.
Listen: www.possible.fm
Daily and weekly AI news
Use one fast news feed to keep up with the firehose without turning your entire listening queue into news.
The AI Daily Brief
Hosted by Nathaniel Whittemore (NLW). Short, frequent analysis of model releases, enterprise AI, policy, products, and industry developments, designed to compress the news cycle into a manageable daily feed.
Listen: podcasts.apple.com/us/podcast/the-ai-daily-brief-artificial-intelligence-news/id1680633614
Last Week in AI
Hosted by Andrey Kurenkov and Jeremie Harris. A weekly roundup of important AI news, research, product releases, and policy developments with discussion rather than a headline-only format.
Listen: podcasts.apple.com/us/podcast/last-week-in-ai/id1502782720
Best AI podcasts for beginners and business listeners
Not every listener needs a two-hour technical debate. These AI in business podcasts explain AI news and tools in plain language, in shorter episodes, for managers, marketers, founders and anyone new to AI.
Everyday AI
Hosted by Jordan Wilson, a former journalist. A daily livestream and podcast about practical AI at work: new features, tools, news, and plain-language how-to episodes. If you are new to AI, the Start Here series is a useful entry point. Daily on weekdays.
Listen: www.youreverydayai.com
The Artificial Intelligence Show
Hosted by Paul Roetzer and Mike Kaput. A weekly news breakdown for business leaders and marketers. Episodes explain what major AI releases and industry changes can mean for work, teams, and jobs.
Listen: www.marketingaiinstitute.com/podcast-showcase
Mixture of Experts
Hosted by Tim Hwang at IBM. A weekly panel with IBM researchers and engineers discussing AI news, models, infrastructure, and enterprise adoption. It sits between business analysis and technical detail.
Listen: www.ibm.com/think/podcasts/mixture-of-experts
Big Technology Podcast
Hosted by Alex Kantrowitz. Interviews with technology executives and researchers, plus regular news discussions. The show uses plain language and is strong on technology strategy and the business implications of AI.
Listen: www.bigtechnology.com
AI Inside
Hosted by Jason Howell and Jeff Jarvis. A weekly show that explains AI news for a broad audience and looks at how AI changes technology, media, products, work, and daily life.
Listen: www.aiinside.show
How I AI
Hosted by Claire Vo. Short, practical episodes where guests demonstrate how they use AI tools in real work, including product management, engineering, writing, and operations. It is useful for listeners who want workflows they can adapt.
Listen: www.chatprd.ai/how-i-ai
AI safety, policy, and frontier labs
For engineering leaders who want more context on AI governance, alignment, risk and geopolitics.
80,000 Hours Podcast
Long-form interviews on AI safety, alignment, governance, biosecurity, and related policy questions. A good fit when you want more depth than a news recap can provide.
Listen: 80000hours.org/podcast
ChinaTalk
Hosted by Jordan Schneider. Covers US-China technology competition, chips, industrial policy, and AI, adding geopolitical context that complements more engineering-focused shows.
Listen: www.chinatalk.media
Data engineering and AI intersection
For teams where reliable AI depends on pipelines, data platforms, orchestration and production data quality.
Data Engineering Podcast
Hosted by Tobias Macey. A long-running show about data infrastructure with growing overlap between data engineering and AI, including pipelines, orchestration, AI-ready data systems, and production architecture.
Listen: www.dataengineeringpodcast.com
Data Skeptic
Hosted by Kyle Polich. A long-running show that applies critical thinking and the scientific method to data science, statistics, machine learning, and AI claims. It is a useful counterweight to hype.
Super Data Science
Hosted by Jon Krohn. Interviews about machine learning, AI, data careers, and tools. The audience is broad, but many episodes go deep enough to be useful for technical practitioners.
Listen: www.superdatascience.com/podcast
Python and general engineering with heavy AI coverage
These are not all AI-first shows, but they consistently cover AI engineering topics relevant to Python developers and engineering leaders.
The Pragmatic Engineer
Hosted by Gergely Orosz. A software engineering show covering Big Tech and startups, with substantial recent attention to AI coding tools, developer workflows, hiring, and the way AI is changing engineering organizations.
Listen: newsletter.pragmaticengineer.com/podcast
Talk Python To Me
Hosted by Michael Kennedy. A weekly Python podcast with increasing coverage of AI-native Python development, agent frameworks, MCP, Pydantic AI, and the tools Python developers use around modern AI systems.
Listen: talkpython.fm
Python Bytes
Hosted by Michael Kennedy and Brian Okken. A short weekly roundup of Python ecosystem news, including recurring coverage of AI-related libraries, tools, and engineering patterns.
Listen: pythonbytes.fm
Listening stacks by role
The backlog-guilt trap is real. Subscribing to twenty shows is a commitment most people will not keep. Pick three or four that match your current role, then review the stack as your priorities change.
For a research-leaning engineering leader
Dwarkesh Podcast + Machine Learning Street Talk + Interconnects. This mix gives you frontier-lab interviews, technical debate and a strong view of open models, RLHF and post-training.
For an AI engineer shipping to production
Latent Space + Practical AI + MLOps Community Podcast. This stack covers agents, evals, deployment, infrastructure and the operational issues that show up after a prototype becomes a production system.
For a CTO translating AI to the business
Training Data + The Cognitive Revolution + The AI Daily Brief. It combines founder and investor framing, builder interviews and compressed daily news.
For a Python-first engineering organization
The Pragmatic Engineer + Talk Python To Me + Practical AI. The mix covers Python ecosystem changes, AI-native engineering patterns and applied deployment.
For a data-leaning platform team
Data Engineering Podcast + Vanishing Gradients + The TWIML AI Podcast. It covers data infrastructure, agents, evals and the research signal that feeds platform decisions.
For governance, safety, and policy focus
80,000 Hours + Interconnects + ChinaTalk. Use this stack when governance, alignment, open-model policy or geopolitical exposure matters to your role.
If your team builds AI products, these guides go deeper: AI coding assistant statistics for 2026 and the best agentic AI frameworks compared. When you need senior engineers, Uvik Software provides senior AI and ML engineers, agentic AI developers and AI agent development services.
Practical listening tip: 1.5x speed works well for conversational episodes. Drop to 1x for technical discussions involving code, math or architecture. If an episode has been sitting untouched in your queue for two weeks, remove it instead of building queue guilt.
Episode length and frequency
A quick reference for matching shows to the time you actually have.
| Podcast | Avg length | Frequency | Format |
|---|---|---|---|
| Lex Fridman Podcast | 2-4 hrs | Weekly | Interview |
| Dwarkesh Podcast | 60-120 min | Weekly | Interview |
| Machine Learning Street Talk | 90-150 min | Fortnightly | Technical debate |
| The TWIML AI Podcast | About 60 min | Weekly | Research interview |
| Interconnects | 30-60 min | Weekly | Essay + interview |
| Latent Space | About 75 min | Weekly | Engineering interview |
| Practical AI | 50-70 min | Weekly | Applied AI |
| Vanishing Gradients | 60-90 min | Fortnightly | Long-form |
| MLOps Community | 45-75 min | Weekly | Practitioner interview |
| The Cognitive Revolution | 90-120 min | Weekly | Executive interview |
| Eye on A.I. | 30-50 min | Biweekly | Journalist interview |
| No Priors | 35-50 min | Weekly | Founder/investor |
| Training Data | 45-60 min | Weekly | Sequoia-hosted |
| The AI Daily Brief | 15-30 min | Daily | News analysis |
| Last Week in AI | 60-90 min | Weekly | News roundup |
| 80,000 Hours | 90-180 min | Weekly | Policy interview |
| The Pragmatic Engineer | 60-90 min | Weekly | Engineering interview |
| Talk Python To Me | 60-75 min | Weekly | Python + AI |
| Python Bytes | 15-20 min | Weekly | Python headlines |
| Everyday AI | About 40 min | Daily (weekdays) | News and how-to |
| The Artificial Intelligence Show | About 70 min | Weekly | Business news analysis |
| Mixture of Experts | About 40 min | Weekly | Expert panel |
| Big Technology Podcast | About 60 min | Twice weekly | Interview and news |
| AI Inside | About 70 min | Weekly | News discussion |
Where to listen
| App | Standout feature |
|---|---|
| Apple Podcasts | Reliable discovery and recommendations |
| Spotify | Large catalog and growing video podcast support |
| YouTube | Full video for many long-form shows |
| Overcast | Smart Speed silence trimming on iOS |
| Pocket Casts | Strong cross-platform sync |
Most shows on this list distribute across several major platforms. For long-form shows with video, YouTube can be useful when you want the full interview experience, while transcripts are often faster when you need to search for one specific segment or technical point.
Methodology
We first compiled this list in April 2026 and checked it again in September 2026. Each show had to pass five tests:
- Activity: the show published at least one episode in the 90 days before the check.
- Focus: AI is the main topic, or AI is the topic of most recent episodes.
- Signal: the four most recent episodes show technical depth, access to the people who build AI, or practical insight. We cut shows that repeat press releases or generic hype.
- Fit: the show serves a clear role, such as AI engineer, researcher, CTO, founder or beginner.
- Language: English.
Lengths and frequencies come from each show’s recent episodes at the time of the check. If a link has changed, search the show name in your podcast app. The next check is planned for December 2026.
About this resource
This list is maintained by the engineering team at Uvik Software, a Python-first senior engineering partner founded in 2015. Our engineers build AI agents, RAG systems and data platforms for companies in the US, UK and Europe, and we use these shows to keep up with the field. If you want senior Python and AI engineers in your team, talk to us.
You can share this list or link to it from your own resources. Attribution is appreciated.
Changes in this update (September 2026)
- Added a new section with six shows for beginners and business listeners.
- Checked every show for a new episode in the 90 days before the check. Removed: The Gradient Podcast, Deep Papers, How AI Is Built, High Signal, The Robot Brains Podcast, The AI Podcast (NVIDIA), ML Platform Podcast, Pipeline Conversations, The MLOps Podcast, Azeem Azhar’s Exponential View, Generative Now.
- Replaced an unsourced claim about agent capability with the METR measurement.
- Added direct answers to two common questions: the number one AI podcast and the top 3 AI podcasts.
Need senior AI engineers in your team?
Uvik Software has 50+ senior engineers with 7+ years minimum production experience, published rates of $50 to $99 per hour and a 30-day no-cost replacement.
See how we work: How Uvik Software works
