HyperAIHyperAI

Command Palette

Search for a command to run...

AI协作催生高质量数学证明:从错误修复到论文级文本的生成与优化

在进一步提示下,ChatGPT成功调整了原有论证,使其既能处理较大的变量,也能适用于较小的变量,最终产出了一个符合原始问题精神的新结果,相关证明可查阅链接:https://drive.google.com/file/d/1xRw8_o2C8HwmxMDnBR5OJlxXaW7jlYbz/view?usp=sharing。值得注意的是,该证明中存在一些细微错误,但AI工具Aristotle能够自动识别并修复这些漏洞,生成了一份经过Lean验证的正确证明。 随后,第三位参与者再次运行Aristotle,对已有的Lean证明进行简化,得到了一个更紧凑的版本。这一简化后的证明被另一位参与者输入至一段长时间的ChatGPT交互对话中(链接:https://chatgpt.com/share/695e7cbd-605c-8010-809b-ccba75560c76),通过多轮讨论与优化,将其扩展为一篇结构更完整、逻辑更严密的文章。该文章不仅清晰呈现了证明过程,还深入阐述了其与已有文献的关联,并构建了更紧凑的叙事脉络。 最终形成的论文草稿(链接:https://drive.google.com/file/d/1MRQfcHhrYMfMTvlZcMC3zEK7aOrUyHiQ/view?usp=sharing)已显著减少了典型的AI生成文本痕迹,整体语言风格和学术表达已接近可接受的研究论文标准,尽管仍有优化空间。作者本人已在论坛(https://www.erdosproblems.com/forum/thread/728#post-2852)对该文本进行了详细点评,认为其质量已达第四阶段水平(4/5)。

相关链接

<p>Meanwhile, with further prompting, ChatGPT was also able to adapt the argument to handle large ? as well as small ?, thus finally producing a new result in the spirit of the intended question <a href="https://drive.google.com/file/d/1xRw8_o2C8HwmxMDnBR5OJlxXaW7jlYbz/view?usp=sharing" target="_blank" rel="nofollow noopener" translate="no"><span class="invisible">https://</span><span class="ellipsis">drive.google.com/file/d/1xRw8_</span><span class="invisible">o2C8HwmxMDnBR5OJlxXaW7jlYbz/view?usp=sharing</span></a> . Interestingly, the proof contained some minor errors in it, but the AI tool Aristotle was able to automatically repair these gaps and produce a Lean-verified proof.</p><p>At this point, a third particiant ran Aristotle again on the existing Lean proof to provide a shorter version, which a different participant then input into a lengthy back-and-forth ChatGPT session <a href="https://chatgpt.com/share/695e7cbd-605c-8010-809b-ccba75560c76" target="_blank" rel="nofollow noopener" translate="no"><span class="invisible">https://</span><span class="ellipsis">chatgpt.com/share/695e7cbd-605</span><span class="invisible">c-8010-809b-ccba75560c76</span></a> to turn it into a much more fully fleshed article that described not just the proof itself, but its connection with prior literature and with a tighter narrative structure. This resulted in a new writeup of the proof <a href="https://drive.google.com/file/d/1MRQfcHhrYMfMTvlZcMC3zEK7aOrUyHiQ/view?usp=sharing" target="_blank" rel="nofollow noopener" translate="no"><span class="invisible">https://</span><span class="ellipsis">drive.google.com/file/d/1MRQfc</span><span class="invisible">HhrYMfMTvlZcMC3zEK7aOrUyHiQ/view?usp=sharing</span></a> that had less of the feel of a generic AI-produced document, and which I judge to be at a level of writing within ballpark of an acceptable standard for a research paper, although there is still room for further improvement. (I review this text at <a href="https://www.erdosproblems.com/forum/thread/728#post-2852" target="_blank" rel="nofollow noopener" translate="no"><span class="invisible">https://www.</span><span class="ellipsis">erdosproblems.com/forum/thread</span><span class="invisible">/728#post-2852</span></a> ). (4/5)</p>
陶哲轩陶哲轩
AI协作催生高质量数学证明:从错误修复到论文级文本的生成与优化 | 热门资讯 | HyperAI超神经