Transformer 联合发明人 · OpenAI 架构研究负责人Co-inventor of Transformer · Architecture Research at OpenAI
nlp
杜克大学数学与计算机科学学士（1998）。2000年作为最早期工程师之一加入 Google，大幅改进了 Google 搜索的拼写纠错系统。成为现代 AI 最高产的研究者之一：合著 Transformer 论文（2017）、开创混合专家模型（MoE）扩展（2017）、构建聊天机器人 Meena。Google 拒绝公开发布 Meena 后，2021年离开联合创办 Character.AI。2024年8月，Google 以 27 亿美元许可 Character.AI 技术，Shazeer 回归 Google 联合领导 Gemini 开发。入选 TIME 100 AI（2023），当选美国国家工程院院士（2026）。
Earned a BS in Mathematics and Computer Science from Duke University (1998). Joined Google in 2000 as one of the earliest engineers, where he significantly improved Google's search spelling corrector. Co-authored the Transformer paper (2017), worked on Mixture of Experts scaling (2017), and built the chatbot Meena. After Google declined to release Meena publicly, he left in 2021 to co-found Character.AI with Daniel de Freitas. In August 2024, Google licensed Character.AI's technology in a deal worth $2.7B, bringing Shazeer back to Google to co-lead Gemini development. Named to TIME 100 AI (2023) and elected to the National Academy of Engineering (2026). In June 2026 he left Google to join OpenAI as Lead of Architecture Research.