This Week’s News

  1. Oracle’s Larry Ellison and Tesla’s Elon Musk beg Jensen Huang, CEO of Nvidia, for more AI chips for their supercomputing clusters. [Link]

  2. AI is failing to live up to hype around pharmaceutical drug discovery. [Link]

  3. OpenAI previews its latest model, GPT-4o1, a first of its kind ‘reasoning’ LLM. [Link]

  4. The White House is launching a new task force on AI datacenter infrastructure. [Link]

  5. Speaking to an AI Chatbot seems to reduce conspiracy theory beliefs in users. [Link]

  6. Meta scrapes all information from its Australian users to train their AI models. [Link]

  7. New paper reveals that LLMs can generate novel research ideas. [Link]

  8. Apple Computer’s iPhone 16 arrives with AI augmentation. [Link]

  9. The Republic of Ireland opens a GDPR investigation into Google over AI practices. [Link]

  10. JP Morgan Chase, the international investment bank, is rolling out its AI assistant to 140 thousand employees. [Link]

  11. Actor James Earl Jones, who died this week, signed his voice rights over to allow AI to replicate it. [Link]

  12. The government of New Zealand is launching their new chatbot, Gov-GPT, modeled on ChatGPT that is designed to help their citizens interact with government services. [Link]

  13. Weave Robotics introduces Isaac, its first generation home assistant robot. [Link]

A robot standing at the edge of a lake

Reflection: An AI Model that Learns from Its Mistakes—and Sometimes Fails

Reflection-Tuning is a groundbreaking AI technique that enables models to self-evaluate and correct errors. Is this the future of artificial intelligence, or a reflection of its current limitations?