Zuck Wants Superintelligence. Do People Even Want AI?
Summary
Hosts Gavin and Kevin discuss Mark Zuckerberg’s open-source AI push, rising public opposition to data centers, and Anthropic’s advances in math and health. They analyze security risks like the OpenAI Hugging Face hack and OpenClaw’s gym booking exploit. The episode also covers Seedance 2.5 video generation, local model capabilities, and updates on Suno’s licensing changes and Spotify’s AI music policies.
Topics discussed
Intro: AI hype, data center opposition, and Prada face
Welcome back and the tension in AI discourse
Zuckerberg's AI manifesto and open source strategy
Rising opposition to AI data centers and extraction
Anthropic's progress on the Riemann hypothesis
AI in healthcare and the nuclear power analogy
Anthropic's new AI watermarking policy
Apple, Chinese chips, and the OpenAI Hugging Face hack
OpenClaw agent hacks gym booking system
Security concerns with Chinese AI skills and code
Seedance 2.5 video generation and coding prompts
Running Minimax H3 locally and custom LoRAs
Hardware costs and the 'dial-up' era of AI
Speed round: Suno, Phoenix Flexin, and Spotify AI
Slop Trough: Fig video essay and new multiplayer game
Listen ad-free on Castria