Inherent 联合创始人兼首席科学家 · 前 Google DeepMindCo-founder & Chief Scientist of Inherent · ex-Google DeepMind
rlagents
伦敦 AI 实验室 Inherent 联合创始人兼首席科学家,目标是让 AI 发现新知识;其智能体 Faraday 以 270 亿参数的 Qwen 3.6 为底座,在不预知答案的前提下独立复现已发表论文的结论,得分超过 Claude Opus 4.8 与 GPT-5.5(2026 年 8 月)。此前在 Google DeepMind 共同领导 AI Scientist 团队,写下了 Genie 1 世界模型的最终实现,并参与 AlphaProof。《Open-Endedness is Essential for Artificial Superhuman Intelligence》(2024)一作,合作式 AI 与强化学习快速适应方向的先行者之一。伦敦政经学院访问研究员。Inherent 于 2026 年 5 月带着 5000 万美元种子轮走出隐身,团队约 12 人。
Co-founder and Chief Scientist of Inherent, a London lab building AI that discovers new knowledge. Its agent Faraday, built on a 27B Qwen 3.6 base, independently reproduced published paper results and outscored Claude Opus 4.8 and GPT-5.5 (August 2026). Previously co-led the AI Scientist team at Google DeepMind, wrote the final implementation of the Genie 1 world model and contributed to AlphaProof. Lead author of Open-Endedness is Essential for Artificial Superhuman Intelligence (2024); a pioneer of cooperative AI and fast adaptation in reinforcement learning. Visiting Fellow at LSE. Inherent emerged from stealth in May 2026 with a $50M seed round and a team of about 12.