原帖引用 · AI 翻译
@Iris97w 嗯，从现在起，人类只会是核查者了。
@rohanpaul_ai 培养人才的流程也会受冲击——研究生是通过阅读证明来学习这个领域的，而不只是核查证明，哈哈。
原文
@rohanpaul_ai the training pipeline breaks too — grad students learn the field by reading proofs, not just checking them haha
陶哲轩对 AI 做数学的方式有意见，而这与证明是否正确毫无关系。😀
数学家想要 AI 当计算器，而不是竞争对手。
“所以，围绕纳维－斯托克斯方程曾有相当多的研究活动，现在却都有点乱了。最终我们会恢复过来。我想，最终我们会理解那个证明，最终人们也会回到这个领域，但完成它的方式非常不理想。
我认为，有一种负责任的解题方式能够推动一个领域发展；还有一种不负责任的方式，就是这种正在兴起的、某种程度上自动化地对这些问题进行掠夺式开采的方式：你达到了眼前的目标，却以损害这个领域未来的发展为代价。”
——陶哲轩
----
完整视频见“New Scientist”YouTube 频道（链接在评论中）。
原文
Terence Tao has a problem with how AI is doing math, and it has nothing to do with whether the proofs are correct. 😀 Mathematicians wanted AI as a calculator, not a competitor. "So, there was quite a lot of activity around Navier-Stokes, and now it’s all kind of in disarray. Eventually, we will recover. I think eventually we will understand that proof, and eventually people will come back into the field, but the way it was done was very suboptimal. I think there’s a responsible way to solve problems that develops a field, and then there’s this irresponsible way, which is this emerging, sort of automated strip mining of these problems, where you get your immediate goal but at the cost of damaging the future development of the field." - Terence Tao ---- Full video on "New Scientist" YouTube channel, (link in comment)