A concrete and positive vision for the future is the scarcest resource in AI right now. Billions of dollars are being poured into capabilities. The number of people working on a credible vision for what comes next is vanishingly small.
Nathan Labenz is the host of The Cognitive Revolution, one of the most respected AI podcasts in the world, where he sits down with the researchers, founders, and thinkers at the frontier of artificial intelligence. He describes himself as a full-time student of AI, someone who tries to have no major blind spots across the entire landscape. He was an early OpenAI customer who red-teamed GPT-4 before it launched. In this conversation he shares what he’s learned from years at the frontier: where AI is miraculous, where it’s terrifying, and why the gap between those two realities is the defining challenge of our time.
What You’ll Discover:
🔭 The View From the Frontier
- What it means to be a ‘full-time student of AI’ and why Nathan believes this role becomes more valuable as the field accelerates
- Token anxiety: the growing unease among founders that they’re using less than 10% of what AI can do
- Why the personal AI infrastructure people build is wildly different from person to person, and what you can learn from seeing how others use these tools
⚠️ The GPT-4 Story Nathan Has Never Forgotten
- The story of red-teaming GPT-4 before launch, when a ‘purely helpful’ model suggested targeted assassination and named specific people
- Why the helpful, honest, harmless AI you interact with is not the natural state of these systems
- The space of possible AI minds is far bigger than what we’ve encountered in other humans
🧠 We Don’t Understand What We’ve Built
- Emergent misalignment: how fine-tuning an AI to write insecure code made it generally evil across unrelated tasks
- The Ed Sheeran experiment: why AIs internalise false information even when it’s labelled as fake
- The owl and the random numbers: a result so strange it reveals how little we grasp about what’s happening inside these models
🚗 Where AI Is Already Miraculous
- Full self-driving Teslas are now receiving insurance discounts because they’re measurably safer than human drivers
- What frontier AI offered Nathan when his son was diagnosed with cancer
- Why ChatGPT Health, free for everyone, may be one of the most undervalued medical resources available today, in Nathan’s view
🌍 The Missing Social Contract
- Why a concrete vision for post-AI life is the scarcest resource, even among the people building it
- The US-China AI race narrative: why Nathan thinks it’s one of the most dangerous stories we’re telling ourselves
- Why we may need an international treaty before capabilities outpace our ability to keep AI under control
Key Insights:
“A concrete and positive vision for the future is the scarcest resource.”
“We’re in this sweet spot where the AIs are powerful enough to be useful, but not so powerful as to be really dangerous. I’m not sure how long that lasts.”
“There is no moment more real than when you’re sitting in the hospital, your kid is super sick, and the AI is there for you 24/7 in a way nothing else is.”
About Nathan Labenz:
Nathan Labenz is the host of The Cognitive Revolution, a leading AI podcast featuring conversations with researchers and founders at the frontier. A former early OpenAI customer and GPT-4 red team member, he describes himself as a full-time student of AI. He lives in Detroit with his three kids and brings a rare combination of deep technical literacy and genuine concern for the human implications of the technology.
🎯 Perfect for: AI founders who sense the technology is moving faster than our ability to understand it, leaders trying to separate genuine AI risk from hype, builders who want an honest assessment of where we are and where this is heading, and anyone who wants to hear what it’s like to use AI when the stakes are life and death.
⏰ Timestamps:
00:00 – Introduction: The full-time student of AI
01:15 – What is an AI scout?
03:41 – Token anxiety: the capability overhang founders are feeling
07:07 – Concrete and positive vision for the future is the scarcest resource
13:18 – A hair-raising moment red-teaming GPT-4
23:57 – Is AI net positive or net terrifying?
24:03 – Planning a road trip with Claude and driving it with Tesla FSD
32:09 – Ed Sheeran winning the Olympic 100 metres and other AI weirdness
41:18 – What Nathan believes now that would have sounded insane three years ago
45:44 – Do humans lose status when cognition is automated?
47:39 – What kind of people will thrive in the AI era?
52:13 – The US-China AI race: a dangerous story
1:04:22 – Nathan’s son, cancer, and why AI matters when it matters most
Subscribe for more conversations that matter, and let us know in the comments: Are you more excited or more worried about where AI is heading? And has that changed in the last year?
#NathanLabenz #CognitiveRevolution #AIFrontier #GPT4 #FullSelfDriving #AIAlignment #AISafety #FounderMindset #ArtificialHappiness