"Next-token predictor" is the wrong mental model for LLMs

Related Stories

• In response to the "wiki incident", OpenAI says it is working on a framework for reporting misalignment incidents during training, evaluation, and deployment
• Q&A with Kai-Fu Lee on China's open-model advantage, AI's impact on jobs, US export restrictions on chips, Chinese AI's role in developing countries, and more
• Scan.com, which uses AI to match patient referrals with imaging centers by availability, price, and specialty, raised a $220M Series C, including $90M in equity
• Portal by Spotify cut my Claude Code token usage by 90%
• Sources: the US and China will discuss AI safety risks during talks planned for mid-September, with Treasury Secretary Scott Bessent leading the US side