Episode 3: Tristan Harris

Video Description

Full audio of this episode here: https://tobias.substack.com/ Tristan Harris starred in (and co-produced) the Netflix documentary The Social Dilemma. He is the founder of the Center for Humane Technology, an ex-Google ethicist, and one of the leading voices warning about how runaway AI might destabilize society. I’ve known Tristan for a long time, and this is one of the best conversations we’ve ever had—public or private. I press him on major risk scenarios, what we can expect AI labs and legislators to do in the face of AGI, and what he thinks can actually be done right now to ensure these systems stay maximally beneficial to humanity. I left this conversation a bit more hopeful understanding what kinds of solutions are available. I hope you do too. Find Tristan's work: http://humantech.com Subscribe here for new episodes: http://tobias.substack.com/podcast My book, Outrage Machine: https://www.outragemachine.org/ 00:00 – Intro 0:40 – Why this conversation matters 1:11 – The economy just flipped (stocks ↑, jobs ↓) 2:29 – Incentives driving the AI race 3:16 – What AI can already do today 4:03 – Shortcuts, slop, and market dominance 5:02 – “Country of geniuses”: labor gets replaced 6:00 – Business models vs. AGI ambition 8:02 – From clickbait to “chatbait” 9:18 – Safety talk vs. engagement DNA 10:30 – What a human-benefiting AI would look like 11:54 – Who captures the gains? 13:47 – Help for juniors, drag on experts 15:04 – Safety movement as accelerator 16:54 – Releasing uncontrollable tech 18:33 – Anthropic’s safety progress & limits 20:06 – Racing to the top (for safety) 22:35 – Capability ↑, controllability ≠ 24:07 – Positive ∞ and negative ∞ in one tool 25:16 – The chimp metaphor (humility check) 27:01 – Narrow, non-anthropomorphic AI (tutors/therapy) 28:34 – Metrics for healthy human–AI use 33:52 – Attachment risks: the “oracle parent” 35:16 – “Humane evals” for long-term effects 35:35 – China’s applied AI vs. U.S. AGI chase 41:15 – Pragmatic policy: liability, testing, kids’ guards 42:02 – US–China guardrails are possible 46:34 – Governance vs. train-wreck governance 50:02 – Deepfakes and collapse of shared reality 52:10 – A creator economy of trust 54:04 – Montreal Protocol as coordination blueprint 55:44 – “Inevitability” is a spell 59:14 – What happens if we do nothing 1:01:40 – What AI leaders should actually do 1:03:15 – Where to help & learn more 52:10 – The Creator Economy of Trust 53:00 – Trainwrecks as Coordinating Events 54:04 – We’ve Solved Global Threats Before (e.g., Ozone) 55:21 – A Different Innovation Environment is Possible 55:44 – Is This Path Really Inevitable? 57:02 – The Spell of Inevitability 59:14 – 13% Job Loss & Runaway Incentives 1:00:15 – Lessons from Social Media: Can We Do Better? 1:01:03 – A Humbling Species-Wide Reckoning 1:01:40 – Final Question: What Should AI Leaders Do? 1:03:15 – Closing Thoughts & Where to Learn More