GPT-6 Astra 登顶 ErdosBench，OpenAI 称未优先优化数学
原标题：GPT-6 Astra gives mathematicians a breather, and OpenAI says that's by design
AI 摘要
OpenAI 发布 GPT-6 Astra，在 ulam.ai 的 ErdosBench 上以 3.23 分排名第一，解决了 226 道开放数学问题中的 106 道，其中 43 道完全解决、27 道被证伪。OpenAI 首席科学家 Jakub Pachocki 表示公司并未优先优化数学能力，而是将资源投入递归自我改进（RSI）和自动化对齐研究，因此 Astra 的数学表现是其他优先级的副产品。数学家 Terence Tao 则警告，若 AI 产出证明的速度超过人类验证速度，数学领域可能从证明稀缺转向证明过载，核心任务将从解题变为判断哪些结果真正重要。
正文节选
GPT-6 Astra gives mathematicians a breather, and OpenAI says that's by design OpenAI's GPT-6 Astra tops the ErdosBench for open math problems, even though chief scientist Jakub Pachocki says the company didn't prioritize math. That says something about where "AGI" actually stands. Mathematicians can take a brief (likely very brief) breather after Astra's release. The field has been grappling with existential questions lately, and while the new model does push math research forward, it's not movi