Transformer founder Noam Shazeer bids farewell to Google and joins OpenAI once again According to Beating monitoring, Noam Shazeer, a core figure in Google's AI and head of Gemini model technology, has left Google again and officially joined competitor OpenAI. OpenAI announced to its employees on Wednesday that Shazeer will focus on finding new underlying architectures for large models and driving the evolution of Transformer architecture after joining. Shazeer is one of the co authors of the groundbreaking paper "Attention Is All You Need" by Google in 2017. The Transformer architecture proposed in the paper is the foundation of modern generative AI models such as ChatGPT, Claude, and Gemini. He joined Google in 2000 and was one of its earliest core engineers. In 2021, due to security and brand concerns, Google refused to release the chatbot Meena developed by him and Daniel De Freitas. Shazeer chose to resign and co founded Character.AI. In August 2024, in order to recall Shazeer, Google reached a licensing and technology acquisition agreement with Character.AI worth up to $2.7 billion. As the founder of Character.AI, who holds approximately 30% to 40% equity, Shazeer personally cashed out $750 million to $1 billion from it. After returning to Google, he became Vice President and Gemini Joint Technical Director, mainly responsible for model pre training. However, less than two years after Google spent a huge amount of money to incorporate him, Shazeer once again chose to leave. Shazeer's joining is an important gain for OpenAI in the talent competition, and it also caused the loss of core technology executives from the Google Gemini team. At present, large models are reaching the computing power expansion limit of Transformer architecture. OpenAI has led the original designers of Transformer to search for alternative solutions, with the aim of seizing the initiative in the next generation of large model architecture to cope with competition from Anthropic and Google. [Original link]