Video summary
A warning about AI incentives, catastrophe risk, and why foresight matters
In this conversation, Sam Harris speaks with Tristan Harris about the risks of AI, the limits of techno-optimism, and the importance of foresight. Drawing on lessons from social media and the history of nuclear fear, they discuss why incentives matter, how future harms can be predicted, and why public awareness may be needed to push for guardrails before a crisis forces action.
AI as a ‘devil’s bargain’
Tristan Harris argues that AI should be judged by likely incentives and outcomes, not just its possible benefits.
Lessons from social media
The conversation links today’s AI race to lessons from social media, which Harris says showed how incentives shape harmful results.
Guardrails before catastrophe
Harris frames the choice as either waiting for a major AI disaster or building guardrails before catastrophe hits.
Why some insiders stay upbeat
The episode discusses the tension between fear and optimism, including how informed insiders can still downplay serious risks.
Topics
From social media to AI
Harris explains how social media offered an early lesson in how incentives can produce harmful outcomes, and why that matters for AI.
Foresight versus catastrophe
The discussion centers on whether AI’s risks should be addressed proactively or only after a disaster makes them unavoidable.
Debating AI optimism
The interview explores why some people inside the AI world remain highly optimistic despite the concerns raised by safety advocates.
Sample transcript excerpt
Transcript
Timestamped transcript passages group captions into readable sections, making the documentary easier to scan, cite, and summarize.
ago about the social element and you started this conversation by saying you know we still are living with all those problems. Well let me give you some good news. India and Indonesia 3 weeks ago joined the list of Australia, Spain, Denmark, France in the set of countries that are banning social media for kids
Sign in to view the full timestamped transcript and use it in Crawlora workflows.