OpenAI近月发布了数百项数学证明，成功解决了困扰数学家数十年甚至数百年的难题。1这一成就在数学界引发了复杂的情感反应。
2022年菲尔兹奖得主Hugo Duminil-Copin表示深感伤心，他说："周二晚上，我哭了，眼泪把我女儿吵醒了...他们摧毁了我的研究领域。"1康奈尔大学的Steven Strogatz也在近期接受采访时表示："我在几周前和《Wired》交谈时真的哭了。"1这些学者的担忧不仅源于问题的解决，更在于对科学本质的反思。他们认为AI虽然成功解决了问题，但中断了科学作为一种代际对话和人类连结的本质。1
中国医学住院医生Shanmu Jin使用ChatGPT解决了Crouzeix猜想，这个问题曾困扰职业数学家数十年。1这一事件具体说明了AI在数学领域快速突破的能力。然而，MIT计算机科学家Regina Barzilay在被问及是否理解自己AI模型在做什么时回答："不理解"，并反问："我们为什么要将一个救命的模型局限于人类能理解的范围呢？"1这种观点与传统数学家的立场形成对比，揭示了自动化知识搜索与人类理解之间的深层张力。1
OpenAI has recently released hundreds of mathematical proofs that solve long-standing problems that have troubled mathematicians for decades or even centuries, prompting complex reactions from the mathematics community.1 Among those responding with concern are Hugo Duminil-Copin, winner of the 2022 Fields Medal, and Steven Strogatz of Cornell University, both expressing deep distress at the implications of artificial intelligence solving these problems.1
Duminil-Copin described his emotional response in stark terms, stating "Tuesday night, I cried and it woke up my daughter... They have annihilated my field of research."1 Strogatz similarly recalled his reaction, noting "I literally cried when I spoke to Wired a few weeks ago."1 Their concern centers on what they perceive as a fundamental disruption to mathematics as a form of intergenerational dialogue and human connection.1 The emotional response reflects anxiety about whether AI-generated knowledge, though mathematically valid, undermines the human understanding and relational aspects that have traditionally sustained scientific inquiry.1
A concrete example of the technology's capability came when Shanmu Jin, a medical resident in China, used ChatGPT to resolve the Crouzeix conjecture—a problem that had confounded professional mathematicians for decades.1 This achievement underscores both the power and the source of unease: while AI can solve problems humans cannot, observers question whether such solutions advance genuine understanding.1 Regina Barzilay, an MIT computer scientist, acknowledged this tension when asked directly whether she understood what her own AI model was doing: she answered "No," and posed a counterpoint by asking "Why would we limit a life-saving model to what humans can understand?"1
评论
还没有评论，欢迎留下第一条。