强化学习大神David Silver，离开DeepMind了。 这位在DeepMind待了整整15年的元老级研究员已经出走，创办自己的AI公司Ineffable Intelligence。 根据注册文件显示，这家公司早在2025年11月就已悄然成立，Silver本人于2026年1月16日被正式任命为公司董事。 在正式离职DeepMind前的几个月里，他也一直处于休假状态。 Ineffable Intelligence总部设在伦敦，目前正在积极招募AI研究人才并寻求风险投资。 Google DeepMind的发言人证实了Silver的离职，并对其在职期间的贡献表示感谢。 除了在谷歌 DeepMind 的工作之外，Silver还是伦敦大学学院的教授，他将继续保持这一职务。 15年老兵，DeepMind的"Alpha系列"缔造者 作为强化学习团队的负责人，Silver主导或深度参与了DeepMind几乎所有里程碑式的项目。 他于2010年公司成立之初便加入，彼时DeepMind还只是一个小团队，Silver和Demis Hassabis在剑桥读大学时是老朋友，他们还一同创办过游戏公司Elixir Studios。 ...

知情人士称，Silver 正在伦敦创办一家名为 Ineffable Intelligence 的新公司。该公司目前正在积极招聘人工智能研究人员，并寻求风险投资。 Google DeepMind 已于本月初向员工宣布了 Silver 的离职消息。Silver 在离职前的几个月里一直处于休假状态，并未正式返回 DeepMind 工作岗位。 Google DeepMind 的一位发言人在电子邮件声明中证实了 Silver 离职的信息，表示：「Dave 的贡献是无价的，我们非常感谢他对 Google DeepMind 工 作所做出的贡献。」 编辑 | 泽南 又一位 AI 大佬决定创业，这位更是重量级。 《财富》等媒体本周五报道说，在 Google DeepMind 众多著名突破性研究中发挥关键作用的知名研究员 David Silver 已离开公司，创办了自己的初创公 司。 根据英国公司注册处 Companies House 的文件显示，Ineffable Intelligence 公司成立于 2025 年 11 月，Silver 于今年 1 月 16 日被任命为该公司董 事。 此外，Silver 的个人网页现在将他的 ...

梦晨 发自 凹非寺 量子位 | 公众号 QbitAI 强化学习大神David Silver ，离开DeepMind了。 这位在DeepMind待了整整15年的元老级研究员已经出走，创办自己的AI公司 Ineffable Intelligence 。 根据注册文件显示，这家公司早在2025年11月就已悄然成立，Silver本人于2026年1月16日被正式任命为公司董事。 在正式离职DeepMind前的几个月里，他也一直处于休假状态。 Ineffable Intelligence总部设在伦敦，目前正在积极招募AI研究人才并寻求风险投资。 Google DeepMind的发言人证实了Silver的离职，并对其在职期间的贡献表示感谢。 除了在谷歌 DeepMind 的工作之外，Silver还是伦敦大学学院的教授，他将继续保持这一职务。 他于2010年公司成立之初便加入，彼时DeepMind还只是一个小团队，Silver和Demis Hassabis在剑桥读大学时是老朋友，他们还一同创办 过游戏公司Elixir Studios。 2016年，他领导开发的AlphaGo击败围棋世界冠军李世石，成为AI发展史上的标志性事件 ...

Core Insights - Google DeepMind's AlphaFold has significantly impacted protein structure prediction, driving advancements in scientific research over the past five years [1][4] - AI is reshaping scientific research, particularly in life sciences and biomedicine, due to rich data availability and urgent societal needs [1][3] Group 1: AI in Scientific Research - AI models and tools have achieved breakthroughs in basic research, including protein structure prediction and the discovery of new biological pathways [1][3] - The paradigm of "foundation models + research agents + autonomous laboratories" is emerging in AI-driven scientific research [3][13] Group 2: Advancements in Biology - DeepMind's AlphaFold has solved the protein structure prediction problem, earning the 2024 Nobel Prize in Chemistry and establishing itself as a digital infrastructure for modern biology [4] - The C2S-Scale model, developed by Google and Yale University, has generated new hypotheses about cancer cell behavior, showcasing AI's potential in formulating original scientific hypotheses [8] Group 3: AI in Drug Development - AI-assisted pathology detection has expanded to new disease scenarios, with the DeepGEM model achieving a prediction accuracy of 78% to 99% for lung cancer gene mutations [10] - The AI-optimized drug MTS-004 has completed Phase III clinical trials, marking a significant milestone in AI-driven drug discovery [10] Group 4: AI in Other Scientific Fields - AI applications in materials science are gaining momentum, with startups like Periodic Labs and CuspAI focusing on discovering new materials [11] - DeepMind's WeatherNext 2 model has surpassed traditional physical models in accuracy and efficiency for weather predictions [5] Group 5: Future of AI in Science - The evolution of scientific intelligence technologies is expected to accelerate, with AI foundational models and robotics enhancing research efficiency [19] - The integration of AI into scientific discovery is anticipated to lead to significant breakthroughs, with predictions of achieving near-relativistic level discoveries by 2028 [19]

Core Insights - Google DeepMind has publicly released the complete technology and training methods behind its new model, AlphaProof, which is designed for mathematical proofs [2][4] - The model utilizes a 3 billion parameter encoder-decoder transformer architecture and incorporates a reinforcement learning environment based on the Lean theorem prover [8][7] Development Process - The AlphaProof team was relatively small, with around 10 core members, and was led by IMO gold medalist Miklós Horváth, who developed a method for creating various problem variants for training [4][5] - Over the course of a year, the team explored various research ideas, integrating successful approaches into the AlphaProof system [5] Training Methodology - AlphaProof transforms the mathematical proof process into a game-like environment, where each mathematical proposition serves as a new game level [7] - The model was pre-trained on approximately 300 billion tokens of code and mathematical text, followed by fine-tuning with around 300,000 manually crafted proofs from the Mathlib library [9][10] - A significant breakthrough was achieved through an automated formalization process that generated about 80 million formalized problems from 1 million natural language math questions [10] Performance at IMO 2024 - AlphaProof demonstrated impressive performance at the 2024 IMO, solving three problems, including the most difficult one, P6, which only 5 out of 609 participants solved completely [15][16] - The model utilized a testing-time reinforcement learning mechanism to generate around 400,000 related problem variants for particularly challenging questions [13][15] Future Directions - Following its success, DeepMind has opened access to AlphaProof for researchers, allowing them to explore its capabilities [19] - While AlphaProof excels in identifying counterexamples and formalizing statements, it faces challenges with custom definitions and relies heavily on the Lean theorem prover [20] - The model's dependency on Lean's evolving environment and the limited availability of unique mathematical problems present ongoing challenges for its broader applicability [20]

Core Insights - DeepMind's AlphaProof achieved a silver medal at the International Mathematical Olympiad (IMO), scoring 28 points, just one point shy of gold, marking a significant milestone in AI's mathematical problem-solving capabilities [3][4][20]. Group 1: AlphaProof's Performance - AlphaProof is the first AI system to earn a medal-level score in a prestigious competition like the IMO, demonstrating a leap in AI's ability to tackle complex mathematical challenges [4][20]. - In the 2024 IMO, AlphaProof solved 4 out of 6 problems, including the most difficult problem, showcasing its advanced problem-solving skills [18][20]. - The performance of AlphaProof is comparable to that of a highly trained international high school student, with only about 10% of human participants achieving gold status [18][20]. Group 2: Technical Mechanisms - AlphaProof combines large language models' intuitive reasoning with reinforcement learning, allowing it to learn from a vast dataset of nearly one million mathematical problems [8][10]. - The system utilizes the Lean formal language for mathematical proofs, ensuring that each step of reasoning is verifiable and free from errors typical of natural language models [6][7][10]. - AlphaProof employs a strategy similar to Monte Carlo tree search, breaking down complex problems into manageable sub-goals, enhancing its problem-solving efficiency [11][17]. Group 3: Limitations and Future Directions - Despite its achievements, AlphaProof's efficiency is limited, taking nearly three days to solve problems that human competitors complete in 4.5 hours, indicating room for improvement in speed and resource utilization [21]. - The AI struggles with certain types of problems, particularly those requiring innovative thinking, highlighting the need for enhanced adaptability and generalization capabilities [21][23]. - Future developments aim to enable AlphaProof to understand natural language problems directly, eliminating the need for manual translation into formal expressions [23][24].

Core Insights - Google DeepMind has publicly released the complete technology and training methods behind its IMO gold medal model, AlphaProof, continuing its tradition of transparency in AI research [1][22]. Group 1: Development and Team Structure - The AlphaProof team was relatively small, typically consisting of about 10 members, with additional personnel joining closer to the IMO competition [3]. - The core breakthrough was attributed to IMO gold medalist Miklós Horváth, who developed a method to create various problem variants for training the AI [3][5]. Group 2: Technical Architecture - AlphaProof employs a 3 billion parameter encoder-decoder transformer model as its "brain," designed to understand the current proof state and output strategies and step estimates for completing proofs [8][9]. - The system transforms the mathematical proof process into a game-like environment, utilizing a reinforcement learning framework based on the Lean theorem prover [6]. Group 3: Training Methodology - The training faced challenges in sourcing sufficient mathematical problems, initially pre-training the model on approximately 300 billion tokens of code and math text [11]. - A specialized translation system was developed to convert natural language math problems into a formal language understood by Lean, generating around 80 million formalized problems from 1 million natural language questions [11][14]. Group 4: Performance and Achievements - AlphaProof demonstrated impressive performance at the 2024 IMO, successfully solving three problems, including the most difficult one, with a training time of 2-3 days per problem [19][20]. - The system's ability to generate related problem variants during testing significantly enhanced its problem-solving capabilities [19][17]. Group 5: Future Directions and Limitations - Following its success, DeepMind has opened access to AlphaProof for researchers, who have reported its strengths in identifying counterexamples and proving complex statements [22][23]. - However, limitations were noted when dealing with custom definitions, indicating a dependency on existing concepts within the Mathlib library [24]. - The reliance on the Lean theorem prover presents challenges due to its evolving nature, which may affect AlphaProof's performance in advanced mathematical fields [24].

Core Insights - Google DeepMind has publicly released the complete technology and training methods behind its IMO gold medal model, AlphaProof, continuing its tradition of transparency in AI research [1][30] - The model utilizes a 3 billion parameter encoder-decoder transformer architecture, which allows it to understand and generate mathematical proofs effectively [12][21] Development Process - The AlphaProof team was relatively small, consisting of about 10 members for most of the development period, with additional members joining closer to the IMO competition [3] - A key breakthrough came from team member Miklós Horváth, who developed a method to create various problem variants for training the AI [4][5] - Over a year, the team explored various research ideas, integrating successful approaches into the AlphaProof system [7] Training Methodology - AlphaProof transforms the mathematical proof process into a game-like environment, where each mathematical proposition serves as a new game level [8] - The system employs a reinforcement learning environment based on the Lean theorem prover, allowing it to suggest strategies and estimate the steps needed to complete proofs [13][14] - The training faced challenges in sourcing sufficient mathematical problems, initially using 300 billion tokens of code and math text for pre-training, followed by fine-tuning with 300,000 manually crafted proofs [16][21] - A significant innovation was the automatic formalization process, which translated natural language math problems into a format understandable by Lean, generating around 80 million formalized problems from 1 million natural language questions [16][21] Performance at IMO - AlphaProof's performance at the 2024 IMO was remarkable, successfully solving three problems, including the most difficult one, despite requiring 2-3 days of computation for each problem [26][28] - The system's ability to generate related problem variants during the competition was crucial for its success [26][27] Future Directions - Following its success, DeepMind has opened AlphaProof's capabilities to the scientific community, allowing researchers to apply for access [30] - Researchers have noted AlphaProof's strength in identifying counterexamples and its limitations when faced with custom definitions in proofs [31][33] - The reliance on the Lean theorem prover presents challenges due to its evolving nature, which can affect AlphaProof's performance in more mature mathematical domains [35] - The limited availability of unique mathematical problems poses a challenge for the AI's generalization capabilities, highlighting the need for further development in generating its own training problems [36]

Core Insights - DeepMind, a subsidiary of Google, has developed an AI system named AlphaProof that can prove complex mathematical theories, enhancing the process of mathematical problem-solving [1][2] - AlphaProof demonstrated its capabilities by solving 4 out of 6 problems in the International Mathematical Olympiad, achieving a score equivalent to a silver medal [2] Group 1: AI System Development - The AI system, AlphaProof, is designed to generate verifiable proofs in a formal mathematical software environment, addressing challenges faced by traditional language models [1] - The system utilizes reinforcement learning to formalize and find proof methods for 80 million propositions, outperforming previous advanced AI systems in mathematical competitions [1] Group 2: Performance and Limitations - In the International Mathematical Olympiad, AlphaProof, in collaboration with another system called AlphaGeometry, successfully solved a significant portion of the competition's complex problems [2] - Despite its impressive performance, experts noted that AlphaProof has limitations in solving other forms of difficult problems, suggesting this as a future research direction [2]

AI Advancement - Google DeepMind's AlphaProof achieved silver medal level performance at the International Math Olympiad last year [1] - Nature is publishing the methodology behind AlphaProof [1]