In this episode of Hidden Layers: Decoded, Ron Green, Dr. ZZ Si, and Michael Wharton explore the latest AI breakthroughs, including Sakana AI’s biologically-inspired “Continuous Thought Machines,” the self-taught coding model Absolute Zero, and Salesforce’s unified vision-language system BLIP3-o. They discuss the growing importance of reinforcement learning in a data-constrained world, Google’s diffusion-based language and video models, and Anthropic’s industry-leading interpretability efforts. The team also covers Apple’s AI missteps and a new study revealing why single, well-structured prompts outperform long chat sessions. Throughout, they reflect on alignment risks, emergent reasoning, and the changing shape of model development and training strategy.
Hidden Layers: AI and the People Behind It