Workflow
AI+数学
icon
Search documents
11位顶尖数学家发了篇没结果的论文,陶哲轩推荐都关注一下
猿大侠· 2026-02-11 04:11
一水 发自 凹非寺 量子位 | 公众号 QbitAI 获陶哲轩转发,arXiv上的一篇新论文正在引起巨大关注! 挤进前排后发现,原来这是一项 由11位全球顶尖数学家发起的AI实验 —— 让AI在规定期限内,解决他们各自在真实研究过程中产生的10道"研究级"难题,以此探索"AI+数学"的能力边界。 而且走的还是高斯时代的路子——人类先证明出来,但先不公布答案和过程,等到了合适时间再公开,避免AI偷偷看答案。 以前这是一项为保护数学家证明自己优先解决某道问题的做法,而在AI时代却有了新玩法。 在陶哲轩看来,这项实验非常有意思: 当前"一次性"AI提示似乎难以解决这些问题,但它们已被人类领域专家攻克。可以预见,配备AI工具的其他领域专家也能解决其中相当 一部分。 这些问题的技术门槛相当高,非领域专家难以验证AI生成的任何输出结果 。 因此在我看来,要让非专家解决其中任何一个问题都极具挑战性——当然,意外惊喜也并非不可能。在截止期限前,这项实验能否产生 任何显著成果,将十分值得关注。 解完10道数学题,然后…藏起证明过程 概括而言,通过提出一套名为First Proof的实验方案,这群数学家想做一件事—— 检验当前A ...
11位顶尖数学家发了篇没结果的论文,陶哲轩推荐都关注一下
量子位· 2026-02-08 04:46
一水 发自 凹非寺 量子位 | 公众号 QbitAI 获陶哲轩转发,arXiv上的一篇新论文正在引起巨大关注! 挤进前排后发现,原来这是一项 由11位全球顶尖数学家发起的AI实验 —— 让AI在规定期限内,解决他们各自在真实研究过程中产生的10道"研究级"难题,以此探索"AI+数学"的能力边界。 而且走的还是高斯时代的路子——人类先证明出来,但先不公布答案和过程,等到了合适时间再公开,避免AI偷偷看答案。 以前这是一项为保护数学家证明自己优先解决某道问题的做法,而在AI时代却有了新玩法。 在陶哲轩看来,这项实验非常有意思: 当前"一次性"AI提示似乎难以解决这些问题,但它们已被人类领域专家攻克。可以预见,配备AI工具的其他领域专家也能解决其中相当 一部分。 这些问题的技术门槛相当高,非领域专家难以验证AI生成的任何输出结果 。 因此在我看来,要让非专家解决其中任何一个问题都极具挑战性——当然,意外惊喜也并非不可能。在截止期限前,这项实验能否产生 任何显著成果,将十分值得关注。 好好好,既然老陶如此安利了,咱这就开扒完整实验过程(doge)。 解完10道数学题,然后…藏起证明过程 概括而言,通过提出一套名为Fi ...
陶哲轩敲警钟,谷歌DeepMind联手五大神殿,用AI向世纪难题宣战
3 6 Ke· 2025-10-30 04:12
Core Insights - Google DeepMind has launched the "AI Empowered Mathematics Program," collaborating with five top global institutions to leverage AI in solving complex mathematical problems [1][2][6] - The initiative aims to discover new mathematical challenges that can benefit from AI, build necessary infrastructure, and accelerate scientific discoveries [6][8] - Concerns have been raised by mathematician Terence Tao regarding the potential misuse of AI in mathematical research, emphasizing the need for responsible use and transparency [2][20] Group 1 - The five collaborating institutions include Imperial College London, Princeton Institute for Advanced Study, Institut des Hautes Études Scientifiques, Simons Institute for the Theory of Computing, and Tata Institute for Fundamental Research [2][6] - The program will be funded by Google.org and will utilize advanced technologies from Google DeepMind [8] - Recent advancements in AI, such as AlphaEvolve and Gemini models, have shown significant progress in solving mathematical problems, including achieving gold medal-level performance in competitions [11][14] Group 2 - AlphaEvolve has provided optimal solutions for 20% of 50 public mathematical problems, including a new efficient matrix multiplication method that broke a 50-year-old record [14][16] - The initiative aims to ensure the rigor of mathematical research while paving the way for the integration of AI and mathematics [5][6] - Terence Tao has proposed a set of guidelines for the responsible use of AI in research papers, including clear declarations of AI usage and discussions on potential risks [23][26]
WAIC 2025|叩响“AI+数学”之问,普陀探寻交融新篇章
Xin Hua Cai Jing· 2025-07-27 05:05
Core Insights - The forum "Mathematical Boundaries and Fundamental Reconstruction of Artificial Intelligence" was held in Shanghai, focusing on the relationship between AI and mathematics, attracting experts from various prestigious institutions [1][2] - The integration of AI and mathematics is becoming increasingly significant, with AI systems like AlphaGeometry demonstrating exceptional capabilities in solving complex mathematical problems [1][2] - The collaboration between AI and mathematics is expected to drive advancements in both fields, with AI helping to address unresolved mathematical challenges while also benefiting from mathematical breakthroughs [2] Group 1 - The forum featured prominent mathematicians, including Professor Shing-Tung Yau, who presented a special problem for AI models to solve, showcasing AI's reasoning capabilities [2] - Experts emphasized the importance of foundational research and original innovation for the advancement of AI in China, highlighting the need for strong theoretical underpinnings [2][3] - The establishment of partnerships between international and local universities symbolizes the collaboration between mathematics and AI, fostering research opportunities [3] Group 2 - The Pudong District is focusing on enhancing innovation in technology and industry, aiming to leverage top-tier technology to strengthen industrial development [4] - Shanghai is actively promoting breakthroughs in mathematical foundations to accelerate AI innovation, aiming to create a comprehensive innovation ecosystem [5]
流星或太阳!广州女孩要卖“数学大脑”给华尔街|热财经
Sou Hu Cai Jing· 2025-06-22 04:53
Core Viewpoint - A young Chinese female PhD, Hong Letong, has founded an AI startup, Axiom Quant, in Silicon Valley, which is reportedly valued at $300-500 million despite having no product or users yet [1][11]. Company Overview - Axiom Quant aims to raise $50 million in funding, with Boston Ventures potentially leading the investment round [1]. - The company is actively recruiting talent to build a team for future research and development [1][11]. Founder Background - Hong Letong, a 25-year-old from Guangzhou, has an impressive academic background, having completed dual degrees in Mathematics and Physics at MIT in just three years and published nine academic papers [6][8]. - She has also been awarded the Alice T. Schafer Mathematics Prize and is one of the few Chinese Rhodes Scholars, further enhancing her credibility in the field [8][11]. Market Positioning - Axiom Quant focuses on providing AI solutions for quantitative finance, targeting hedge funds and quantitative traders [10]. - The startup plans to utilize a "mathematics as a service" model to address complex mathematical proof challenges, aiming to create efficient quantitative models for financial institutions [10][11]. Industry Context - The rise of young entrepreneurs in the AI sector, particularly those with strong academic credentials, indicates a shift in the technology landscape [12][14]. - The intersection of mathematics and AI in quantitative finance presents significant long-term potential, addressing the need for high-efficiency and high-reliability quantitative models in the finance industry [11].
一位广州00后女生爆红
投资界· 2025-06-07 07:29
作者 I 王露 报道 I 投资界PEdaily " 00后们开始起跑了。"本周,一个有点陌生的名字——洪乐潼,密集出现在AI投资人的交流里。 图片来源:麻省理工学院 在学界崭露头角是2 021年底,洪乐潼成功拿到了牛津大学罗德奖学金,成为仅有的四名中国获奖者之一。 这源自于The I nf orma ti on的报道,0产品0客户,甚至还没有正式官网的AI初创公司Axi om正在进行500 0万美元融资,估值约3亿到5 亿美元。有意思的是,公司 将 用AI解决数学难题,与今天的高考十分应景。 消息一出,立刻在硅谷和国内创投圈刷屏,背后创始人一夜爆火——缔造Axi om的正是00后学霸洪乐潼(Ca ri na Hong)。 在广州出生长大,洪乐潼曾就读于著名的华南师大附中,多次拿下奥数竞赛奖牌。后考入麻省理工学院,硕士毕业于牛津大学。如今 20出头的她,正在斯坦福大学攻读数学和法学博士双学位。目前,洪乐潼在社交平台回应融资消息"不实"。但可以确定的是,Axi om 正在招聘AI数学人才。 悄然间,00后创始人正在登上AI舞台。 一位00后女学霸爆红 来自广州 公开资料显示,洪乐潼是位00后,出生于广州市。父母是 ...