I've seen this sentiment from several mathematicians, and it is understandable, but I feel it is also dangerously wrong. I say this because I think mathematicians need to adapt hard and fast, instead of wishing that the disruption from AI will somehow go away. There is a
原文
Tony Feng is a mathematics professor at UC Berkeley, and he tells us: « In the past, the path to a math Ph.D. cultivated resilience, resourcefulness, critical thinking, and a healthy skepticism (…) but suddenly you can produce a passable Ph.D. thesis with the push of a button. » Obviously, it is not just in mathematics. Obviously. Want to write a PhD in sociology, education, computer science, or mechanical engineering? Press of a button. Let us go back in time. I started university before the web. At the time, you had to go to a physical library to do research. You would wander through the (paper!) journals, find something interesting, and either read it there or make a copy or borrow it. When I finished my PhD, I had stashes of journal articles, categorized in some way. I had collected them over years. I did not have a search function, and certainly not an AI helper. So you sort of had to memorize the important papers. Working alone, with little help from your thesis supervisors, was not great. That was an era where the quality of your thesis supervisor mattered a great deal. If they knew the literature of a domain like the back of their hand, they could really direct you to read the proper work. It made a huge difference. Then we got the web, and search engines. At first, I would just dump a lot of PDFs on my disk and search through them for keywords. Then we got Google Scholar. What a breakthrough! A full index of the entire scientific literature. It was like having a friend with an encyclopedic knowledge of the literature with you at all times. At that point, if you were a lone fellow in some corner of the world and somehow had library access to the PDFs, you could reasonably, on your own, pick a topic and find out what people had written about it, not in years, but in days or maybe hours. All along, the number of people writing PhD theses went up. in the United States it rose from about 36,000 in 1990 to over 58,000 in 2024, while in China it exploded from just a few thousand in the early 1990s to roughly 88,000 per year today. Then it became much easier to share code, and then data. Even in the physical world, research became somewhat easier to automate. Drones can fly over a forest, taking pictures. We have robots that can conduct chemical analysis and experiments. And then, the biggy: we got agentic AI. I could, today, write a PhD thesis generator. Enter a topic, and a PDF that looks like what would have been accepted for a PhD in 2015 will come out. Consider that most PhD theses were never good. How often do you rush to read a PhD thesis? The vast majority of them are painful to read. You learn little if anything. So the bar is not as high as people imagine. And that’s where we are today. So people tell me: “Yeah, but to get the PhD, you have to present it live and answer questions.” Yeah. About that. For social reasons, it is highly uncommon for people to fail the oral defense of their thesis. Some people are just terrible at it, naturally. They cannot think well on their feet. They get confused and answer the wrong question. I was at an oral defense a few months ago. My older colleague kept raving about the extensive review of the literature. He kept saying that he had never seen such a young person produce such an extensive review in so little time. This same colleague is proud to say that he opposes ChatGPT and friends. Would I tell him? No, I kept my mouth shut. I am never rude on purpose. To sum it up, the very foundation of modern academia has been blown to bits. Not shaken. Wiped out.