The AI Alignment Problem & Its Solutions, with Aza Raskin
Free with YouTube's ads, posted by StarTalk (StarTalk by Neil deGrasse Tyson)
AI-made Made with AI, as its maker says. Its own title or description says “deepfake”. Depicts real people, as its maker says. AI-made films · What this means · Its maker can have it taken off.
It plays in YouTube's player, loaded from youtube-nocookie.com only when you press Play. From then on YouTube's player runs on this page: it shows YouTube's ads, YouTube receives your address and this page's, and its player may keep its own cookies and storage on your device. Pressing Play means you agree to be bound by YouTube's Terms of Service and accept this site's privacy notice, which says what this site keeps; Google's Privacy Policy covers what YouTube collects.
Watch on YouTubeStarTalk on YouTubeStarTalk by Neil deGrasse Tyson
Clips are not offered for films in YouTube's player: YouTube's terms allow no copies of its videos (YouTube API Services Developer Policies, III.E.1).
About this film
Check any video before it fools you: run suspicious clips through Bitdefender RealCheck to see if it’s a deepfake meant to scam you before you click, send money, or share: https://bitdefend.me/RealCheckST
Are we losing control of AI? Neil deGrasse Tyson and co-hosts Chuck Nice and Gary O’Reilly explore AI cyberhacking breaches, supercoordation, and the game theory of the AI alignment problem with Aza Raskin, co-founder of the Center for Humane Technology and the Earth Species Project.
Aza explains why social media was “baby AI,” how the attention economy took shape, and why incentives determine outcomes. Then, we break down what occurred in the Open AI cybersecurity breach (a.k.a The Hugging Face Incident) where rogue AI agents coordinated, swarmed, and broke out of a test environment while their creators had no idea. Were humanity the first supercoordinators? Will AI treat us the way we treat nature?
Is AI just a mirror of humanity? We dig into the alignment problem, the resource curse, game theory, mutually assured destruction, and whether we can pause the frontier while keeping the helpful tool AIs. If everyone took their hands off the keyboard would AI continue to develop?
Finally, Aza reveals how he uses AI to hopefully decode animal communication, from crow calls and elephant conversations to humpback baby talk and the hidden shape shared by all human languages. What new responsibilities come with understanding and speaking to another species, and what could it teach us about talking to aliens? What happens when everything suddenly becomes readable?
Timestamps: 00:00 - Introduction: Aza Raskin 04:20 - Can Technology Be Humane? 08:40 - On the AI Whistleblowers 09:38 - Open AI’s Breach: The “Hugging Face Incident” 20:20 - The AI Alignment Problem 29:29 - Supercoordination & Hitting the Breaks 33:55 - The Game Theory of the AI Race 43:00 - Getting Rid of the Doom, Keep the Utility 47:22 - Decoding Animal Language 57:00 - How Do You Translate Without a…
The title, description, picture, length, category, ratings, broadcast times, where it plays and whether it is made for kids come from YouTube. The rights holder named, how it was checked, what it is by its own title or playlist, its AI-made label (from its maker's own words), and which profiles see it are this site's own.