In this episode of the Last Week in AI podcast, Andrey Kurenkov is joined by Michelle Lee to discuss the latest AI news. The topics include OpenAI's upgraded Codex, Google integrating Gemini into Chrome, Anthropic's new spreadsheet and PDF creation feature, and Luma's new video model. They also cover OpenAI's transition to for-profit status, Microsoft's partnership with Anthropic, funding for humanoid robotics companies like Figure.ai and Unitree, Tesla's robotaxi testing and accidents, Amazon's Zoox robotaxi launch in Las Vegas, and funding rounds for Repl.it and Perplexity. Additionally, they discuss K2Think, a parameter-efficient reasoning model, LocalBench, a benchmark for long context language models in software engineering, self-improving embodied foundation models, physics foundation models, and embodied navigation foundation models. Finally, they touch on policy and safety issues, including Anthropic's endorsement of California's AI safety bill and lawsuits against Midjourney and Google over copyright violations.
Part 1: AI Tools and Models
Part 2: Robotics and Funding
Part 3: Research and Benchmarks
Part 4: Policy, Safety, and Conclusion
Sign in to continue reading, translating and more.
Open full episode in Podwise