AI Inside Gemini 3 is Here
17 snips
Nov 19, 2025 Discover the groundbreaking features of Google’s Gemini 3, which integrates text, video, and audio for a seamless user experience. The discussion also dives into Microsoft’s ambitious AI Copilot and its implications for Windows users. Enjoy thrilling debates on the limitations of LLMs and predictions about the AI bubble, as industry figures weigh in. Plus, learn about Jeff Bezos’s well-funded Project Prometheus focusing on engineering and manufacturing innovations. Tune in for insights on the future of AI and its challenges!
AI Snips
Chapters
Books
Transcript
Episode notes
World Models Versus Scaling Debate
- Yann LeCun argues LLMs are a dead end for AGI because they only predict next tokens and lack embodied world models.
- Adam Brown counters that scaling LLMs can still reach general intelligence by trajectory and compute.
LeCun's Open Source Plea
- Jeff Jarvis met Yann LeCun after a debate and recorded his plea for open-source AI to avoid information capture.
- LeCun warned monopoly control of AI would skew culture, language, and democracy if left to a few firms.
Gemini 3 Expands Multimodal And Agent Tools
- Gemini 3 adds native multimodality and Dynamic View to ingest text, audio, images, and video simultaneously.
- It also offers agentic tools and an IDE-like multi-agent environment called Anti-Gravity for complex development.
