#119 - Open Source GPTs, X.AI, Auto-GPT, China’s Censorship of AI, Fake Drake+The Weeknd Colab
Summary
Hosts Andrei Karpathy and Jeremy Bernstein discuss the rise of open-source AI, including Stability AI’s new models and the Open Assistant project, while analyzing Stability AI’s business struggles. They cover Elon Musk’s X.AI launch, Google merging DeepMind with Brain, and Amazon’s AWS Bedrock expansion. The episode also explores the "Llama Effect," where leaked Meta models spurred a wave of high-quality open-source alternatives like Vicuna.
Topics discussed
Sponsors: OutShift by Cisco and Box
Intro and episode overview
Stability AI's open source models and the open source movement
Stability AI's business model and valuation concerns
Elon Musk's new AI initiative and safety debates
DeepMind becomes Google DeepMind
Amazon AWS expands generative AI efforts
Boston Dynamics' new robot at CES
Open source ChatGPT alternatives and scaling laws
AutoGPT and autonomous AI agents
Lightning round: Black hole images, tennis robot, and consistency models
AI for stroke prediction in hospitals
AI policy roadmaps and China's new regulations
AI existential risk and the talent war in academia
AI-generated art controversy and creativity tests
AI voice cloning, Adobe Firefly, and Meta's animation tool
Listen ad-free on Castria