One study found that math graduate and postdoctoral students tend not to want to specialize in this field because AI is getting better.

The survey was conducted by three researchers from the Massachusetts Institute of Technology (MIT) from September 17 to October 8 with 268 people, including 85% doctoral students and 15% postdoctoral researchers, at nearly 60 institutes and universities in 5 countries.

The results were announced over the weekend on the arXiv archive. Questions include multiple choice and essay answers. Most revolve around AI's ability to quickly solve problems that require years of tinkering.

Specifically, 56% said they were considering changing their pursuit of specializing in math but had not decided yet, 8% had changed, 17% had considered but chose to keep the plan and 19% had not thought about it. Nearly 76% said there was "less and less motivation" to continue pursuing academic mathematics after graduation. Meaning, nearly two-thirds of participants have considered or decided to adjust their career paths in light of AI developments.

The worry is more evident in job prospects. Of the 262 people who responded to a related question, 76.3% said that AI makes it difficult for them to find tenure-track faculty positions at universities. Only 6.9% predicted technology would make it easier for them to find a job.

Tenure-track is a career path towards a tenure-track faculty position, often based on criteria of research, teaching and scientific publication. For graduate students and postdoctoral researchers, this is one of the important career development directions after many years of training.

In the survey's essay section, some even expressed their feelings toward the survey, including disappointment with academic leaders, describing them as "stubborn, weak, silent, and ineffective."

"It would be great if institutions provided guidance instead of empty words on this issue," one survey participant wrote.

“We are lacking senior leadership from across the academic community,” said another.

Even though they are dissatisfied with AI, many people still use tools to help in many stages of research, from finding documents to checking proof. Of the 227 people who answered about the purpose of use, 85% said they used AI to search for documents, 68% used it to edit articles and fix typing errors, and 58% used it to check mathematical errors. About 54% use AI to prove specific propositions, 52% to find ideas, 50% to help create or fix errors in source code, 19% to create content for research articles, and 13% to create complete solutions from start to finish.

Using AI does not mean participants are satisfied with the technology. Of the 230 people who talked about their experiences, 54.8% were inclined to dislike, 30.9% were inclined to like. 76.5% of the 264 respondents said they would not be satisfied if AI or a lab using AI solved the problem they were researching, with only 9.8% leaning towards being satisfied.

According to the research team, those surveyed are not a representative sample of the entire mathematical community, but are showing that AI is already starting to impact research motivation, career development opportunities, knowledge creation and the future of mathematics.

"The past two months have been extremely tiring," the researchers wrote about the survey period. "We are forced to face life-or-death questions about the field we love, while feeling extremely insecure about our own future."

The results were published in the context of AI being able to quickly find solutions to a series of difficult problems that experts have not had answers to for many years. In early October, OpenAI published 722 drafts containing 372 mathematical results, claiming to have made progress on hundreds of open problems despite having to withdraw three drafts due to detecting errors, and edit 14 other drafts.

According to Business Insider, many mathematicians criticized the speed and scale of this announcement, saying that the results need to be carefully verified, clearly explained and previous work properly acknowledged. The Mathematical Humanities Association with more than 800 members also issued a protest notice, saying this does not actually promote mathematics forward, and even called for a boycott of OpenAI. In September, 25 mathematicians who received the Fields Medal signed an open letter, saying that the mass "dumping" of mathematical research could harm the field, disrupting the process of research and knowledge transmission. Math prodigy Terence Tao also believes that AI is causing the "romance" of mathematics to end.

Some experts say companies like OpenAI disrupt the work of young mathematicians. "Our entire research program was wiped out because of their indiscriminate dumping," Professor Tristan Buckmaster at New York University told CNBC in early October, after OpenAI announced its math results.

Bao Lam summary

• Sam Altman: The mathematicians turned from despair to acceptance of AI
• Prodigy Terence Tao: AI brings the 'romance' of mathematics to an end
• After mathematics, AI can conquer new heights
• Sam Altman: 'The world should accept the bad things about AI'