What should we tell our students?

Your math research projects could triple in scope, becoming more accessible with AI tools.

A student expressed deep concern to mathematician Álvaro Lozano-Robledo about the future of an academic career in mathematics due to rapid advancements in AI and Large Language Models (LLMs). The student, who had aspirations for a PhD and research, questioned whether the field would remain accessible to passionate individuals or become exclusive to a select few "brightest minds," and whether AI would surpass human capabilities in pure math research within 5-10 years. Lozano-Robledo acknowledges these widespread anxieties within the mathematics community, noting that many students are reconsidering graduate studies. He points out that AI companies control the narrative about LLM capabilities, often highlighting successes in solving complex problems while omitting the significant costs and failures. These companies have corporate interests, particularly with upcoming IPOs, in convincing the public of their technology's immense and imminent potential. The author, a tenured professor, aims to address these student concerns, though he stresses that definitive answers are currently unavailable due to the rapid and opaque nature of AI development.

The narrative surrounding AI's impact on mathematics is being shaped by frontier model companies, whose corporate interests, including upcoming IPOs for companies like Anthropic and OpenAI, incentivize them to promote the idea of imminent "super-human intelligence" in AI. These companies invest heavily in solving well-known mathematical problems to showcase their models' capabilities, but often omit the vast resources spent on failures and the high operational costs, creating a potentially misleading public perception. This corporate-driven narrative creates anxiety among students and mathematicians, who fear obsolescence and a shrinking academic landscape.

Despite the concerning narrative, Lozano-Robledo expresses optimism, viewing the current period as the most exciting in modern mathematical history. He believes that if LLMs are used correctly and sensible academic conduct expectations are established, these tools can significantly accelerate progress in mathematics. He anticipates that the mathematical community will adapt, harness these tools, and that human mathematicians will remain central to future achievements, with the day-to-day work of a mathematician largely unchanged but with an expanded research horizon and more project opportunities for students.