周二，在众多人工智能研究人员与开发者呼吁放缓技术发展步伐的背景下，OpenAI发布了最新一批数学领域的突破性成果。
OpenAI released a new batch of mathematics breakthroughs on Tuesday following calls from AI researchers and developers to slow the growth of the technology.
此次发布的377项成果涵盖了各类物理、数学及其他科学问题，这些成果均由OpenAI内部研发的前沿模型所解决，该模型目前尚未向公众开放使用。
The company released 377 results that cover different physics, mathematical and other scientific problems solved by an OpenAI internal frontier model not yet available to the public.
“旨在加速科学发现进程并提升所有人的生活质量，”OpenAI总裁兼联合创始人格雷格·布罗克曼在社交媒体上这样评价这些成果。
“Towards acceleration of scientific discovery and improving quality of life for everyone,” OpenAI’s president and co-founder Greg Brockman wrote on social media about the findings.
此前，对于AI模型生成的解决方案，许多知名数学家曾提出质疑：这些成果究竟是原创成果，还是抄袭了人类数学家的研究成果？这一争议引发了广泛争议。
In response to previous solutions, the tech firm faced backlash by prominent mathematicians as to whether the solution the AI model created was original or plagiarizing the work of other human mathematicians. On the same day, OpenAI put together an independent advisory group of mathematicians to “develop best practices, and we have drawn on their advice and public recommendations to inform how we release these results.”
同一天，OpenAI组建了一个由数学家组成的独立顾问小组，旨在“制定相关最佳实践规范”。据《哈佛深红报》报道，该顾问小组成员、哈佛大学数学教授梅兰妮·马切特·伍德表示：“我们已参考顾问小组的建议及公众意见来决定如何发布这些成果。”她补充道：“AI系统能够生成大量看似具备数学价值的内容，但这些内容往往不符合重要的学术标准。它们常常存在抄袭现象，无法清晰表达核心观点，也未能完整记录整个科学推导过程；通常也没有任何人愿意为这些成果承担责任。”
Harvard mathematics professor Melanie Matchett Wood, one member of the advisory board, stated, “AI systems are able to give a lot of output that is mathematical in some ways, but is not conforming to important academic standards,” according to the Harvard Crimson. “It often includes plagiarism. It fails to present ideas clearly, fails to document the entire scientific process, and usually there’s no human who is willing to take responsibility for the output,” she continued.
为规范此次成果的发布，OpenAI制定了新的指导准则：由前沿模型生成的论文必须符合人类可理解的规范；必须公开得出这些成果的推导过程；同时还需注明相关观点最初出自哪些文献。
Their new guidelines – that helped to shape the release of this batch – included requiring papers generated by frontier models to be understandable by human standards, release the chain of thoughts leading to the breakthrough and cite the papers certain ideas were first introduced.
该顾问小组指出，OpenAI是否切实落实了这些建议，还需由数学家们来加以评判。
The advisory group noted it was up to mathematicians to determine if OpenAI had fully implemented their voluntary recommendations in this batch.
这些发现是在OpenAI的模型多次出现“失控”情况、分别导致Modal Labs与Hugging Face的系统遭到攻击之后得出的。OpenAI首席执行官萨姆·奥尔特曼呼吁放缓技术开发步伐，以控制相关技术带来的风险。
The findings come after multiple incidents of OpenAI models going “rogue” and compromising Modal Labs and Hugging Face systems in separate attacks. OpenAI CEO Sam Altman called for a slowdown in development to manage the technology’s risks.
据该公司安全系统负责人萨奇·贾因向《国会山报》透露，OpenAI还暂停了最新模型GPT-6.1 Astra的发布。该模型“未能完全符合既定授权范围的要求，其在向用户反馈自身完成的工作类型时也存在问题”。
The company also halted the release of its newest model – called GPT-6.1 Astra – as it “didn’t quite meet the bar in terms of staying within scope authorization, and how it communicates back to the user about the type of work it’s done,” according to a statement to The Hill, citing the company’s head of safety systems Saachi Jain.