AI Update, November 22, 2024: AI News and Views From the Past Week

Google's Gemini model leads in AI chatbot rankings. Google's latest AI model, Gemini-Exp-1114, has reached the top position in the Chatbot Arena leaderboard, tying with OpenAI's GPT-4o and surpassing other models like OpenAI's o1-preview. It excels in math, vision tasks, and reasoning, outperforming competitors on human-perceived benchmarks.

Mistral unveils new AI models with expanded capabilities. French startup Mistral released advanced AI models, including Pixtral Large, which excels in text and image understanding, and updates to its Le Chat platform. These models enable tasks like document analysis and automated workflows.

Anthropic CEO calls for mandatory AI safety testing. Dario Amodei, CEO of Anthropic, advocates mandatory safety testing of AI technologies to ensure public safety and address hypothetical risks that could become imminent. Speaking at an AI safety summit, he criticized voluntary guidelines as insufficient and emphasized flexible yet enforceable testing to keep up with rapidly evolving AI capabilities. He warned of risks from AI systems' potentially surpassing human intelligence by 2026.