Skip to content
The Joe Rogan ExperienceThe Joe Rogan Experience

Joe Rogan Experience #2076 - Tristan Harris & Aza Razkin

Tristan Harris and Aza Raskin are the co-founders of the Center for Humane Technology and the hosts of its podcast, "Your Undivided Attention." Watch the Center's new film "The A.I. Dilemma" on Youtube.https://www.humanetech.com"The A.I. Dilemma"https://www.youtube.com/watch?v=xoVJKj8lcNQ

Tristan HarrisguestAza RaskinguestJoe Roganhost
Jun 26, 20242h 31mWatch on YouTube ↗

At a glance

WHAT IT’S REALLY ABOUT

AI’s Runaway Race: Power, Profit, and Humanity’s Survival Crossroads

  1. Tristan Harris and Aza Raskin argue that modern AI is repeating and massively amplifying the harms of social media, driven by a profit-and-power race among tech giants and nations. They explain how transformer-based AI systems gain unpredictable, emergent capabilities as they scale, collapsing the distance between dangerous intent and real-world action in areas like bioweapons, cyberattacks, persuasion, and mass deception.
  2. Using examples from social media addiction, infinite scroll, beauty filters, and political polarization, they frame social platforms as “first contact” with AI, a warning of what happens when incentives are misaligned with human well‑being. They warn that open, powerful models and DNA printers could enable small groups or individuals to cause catastrophic harm faster than governments and institutions can respond.
  3. Despite the risks, they emphasize AI’s potential for breakthroughs in medicine, climate, and governance, and argue the core issue is incentives and coordination, not the technology itself. They call for a global shift from a race to deploy offensive AI capabilities to a race to build secure, defense‑dominant, humane systems that strengthen democracy and shared reality.
  4. They close by urging public awareness and political pressure to change AI’s incentive structures, likening this era to a civilizational rite of passage where humanity must mature—embracing our cognitive limits, upgrading our institutions, and taking responsibility for the ‘shadow’ side of our technologies—or face collapse or permanent authoritarian control.

IDEAS WORTH REMEMBERING

5 ideas

Incentives, not features, determine whether technologies help or harm societies.

Social media shows that when systems are optimized for attention and profit, they naturally evolve toward addiction, outrage, polarization, and mental health crises—even if their creators originally intended connection and empowerment.

Modern AI systems gain unpredictable, emergent abilities as they scale.

Transformers trained simply to predict the next word or character spontaneously learn sentiment analysis, chemistry, theory of mind, and code exploitation, meaning creators cannot fully enumerate or anticipate a model’s capabilities before deployment.

AI collapses the distance between dangerous intent and dangerous action.

Unlike static Google search results, interactive AI tutors can iteratively guide users through complex tasks—like building explosives, designing bioweapons, or evading controls—making sophisticated harm accessible to less skilled actors.

Open, powerful AI models are not just insecure—they become insecurable.

Once model weights for a capable system are released (like LLaMA or Falcon), anyone can fine‑tune away safety constraints, proliferate the model globally, and use it as a “teaching tool” to jailbreak more powerful closed systems, with no practical way to revoke or recall it.

AI is amplifying an information and reality crisis that threatens democracy.

As generative models begin producing most online content—voices, faces, text, music, and video—people, institutions, and law enforcement will struggle to distinguish truth from fabrication, overwhelming governance and eroding shared reality.

WORDS WORTH SAVING

5 quotes

If the technology confers power, you're going to start a race. If you do not coordinate, that race will end in tragedy.

Aza Raskin

Social media is kind of a baby AI... It was first contact between humanity and AI, and humanity lost.

Tristan Harris

As fast as everything is moving now, unless we do something, this is the slowest it will move in our lifetimes.

Aza Raskin

We are heading into the largest election cycle the world has ever seen at the exact moment we are deploying the biggest, baddest new technology.

Aza Raskin

It's not about being optimistic or pessimistic. It's about opening your eyes as wide as possible so you can show up and do something about it.

Aza Raskin

Social media as ‘first contact’ with AI and the lesson of misaligned incentivesTransformers, emergent AI capabilities, and runaway scaling dynamicsAI’s dual‑use potential: from tutoring and medicine to bioweapons and cyberattacksOpen models, security risks, and the difficulty of containing powerful AI systemsBiological and information security: DNA printers, bio‑risk, and deepfakesGeopolitical AI arms race: U.S., China, UAE, and game‑theoretic pressuresGovernance, incentives, and possible paths to humane, defense‑dominant AI deployment

High quality AI-generated summary created from speaker-labeled transcript.

Get more out of YouTube videos.

High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.

Add to Chrome