There's never been a better time to go into math

Summary

The video argues that there has never been a better time to enter mathematics, despite uncertainty about AI. The case is not that the future is predictable: the speaker emphasizes that AI progress is rapid and uneven, and that the video’s predictions should be treated cautiously. Rather, mathematics remains valuable in several ways, and the community is unusually open to rethinking how mathematical work is done and recognized.

What mathematicians do

Mathematical work includes more than proving theorems:

• Proving theorems: Establishing that particular claims are true.
• Choosing questions and making conjectures: Judging which problems are worth pursuing.
• Building theories: Creating concepts, definitions, and frameworks that organize many results and make new ways of thinking possible.
• Finding applications: Connecting mathematical ideas to science and engineering.
• Communicating: Explaining results and why they matter, in writing and through teaching or talks.

The speaker sees theory-building as especially important. Individual proofs can matter, but broad frameworks—such as coordinate geometry, group theory, or algebraic geometry—can reshape entire areas of mathematics and eventually enable applications.

Where mathematics gets its value and funding

The video identifies three major sources of value:

1. Applications to science and engineering. Pure mathematics can have unexpected uses, sometimes decades or centuries after it is developed.
2. Transferable problem-solving skills. Mathematical training can help people in fields such as computing, machine learning, finance, cryptography, and AI safety.
3. Aesthetic and personal enjoyment. Many people value the satisfaction of solving a problem or learning a theory that changes how they think.

Funding comes from several sources, including university teaching, public grants, private foundations, and industry. The speaker says teaching is especially important for mathematics. It supports the development of transferable skills and gives students access to the intellectual enjoyment of mathematics, while research can contribute to broader applications.

What if AI becomes better at mathematical work?

The speaker considers increasingly ambitious hypotheticals, rather than claiming to know exactly what will happen:

• If AI is better at proving theorems: Proofs may become easier to generate, but theory-building, applications, education, and the value of practicing mathematics do not automatically disappear. The major change may be to mathematical culture and how credit is assigned.
• If AI is also better at proposing conjectures: Human mathematicians may be able to develop theories faster by examining more results and deciding how they fit together. The value of training people to reason mathematically remains, assuming human problem-solving skills continue to matter.
• If AI can also build theories: People may still need to understand the theories well enough to connect them to science and engineering. The experience of learning and doing mathematics can remain valuable in its own right, though some researchers may miss the experience of being pioneers.
• If AI is better at applications too: In principle, some research funding aimed purely at applications could be directed toward computational systems. But universities and human mathematical institutions may still be supported for education, transferable skills, and the value of mathematics as a human activity.

Across these scenarios, the speaker distinguishes between mathematics as a useful practice for people and the production of mathematical results. Even if machines outperform people at a task, people may still benefit from doing it themselves—like exercising even though machines can lift more weight.

Proofs, understanding, and academic credit

The speaker suggests that proof has often served as a proxy for understanding. If machines can produce proofs, that proxy may become less useful. A proof could be treated more like raw data: potentially valuable, but not necessarily the final achievement.

In this view, the deeper goal is to understand the ideas behind results and develop them into coherent theories. That may require a shift in how the mathematical community measures contributions and awards academic credit. Clear explanations, attribution of earlier human work, and the ability to help others understand and care about the mathematics will matter increasingly.

The speaker also notes that AI-generated explanations may currently be hard to follow, but expects that to improve. Human communication—especially talks and discussion—remains important because people are socially motivated, not just persuaded by clear explanations.

Practical advice for students and researchers

For a student asking what to do next, the video offers these suggestions:

• Reflect on what you love about mathematics. Identify whether your interest is in proving results, building theories, explaining ideas, applying mathematics, or pursuing mathematics for its own sake.
• Choose a direction that fits your interests. Some researchers may use AI tools to explore new mathematical territory; others may choose to develop a distinctly human-centered approach to mathematics.
• Give talks regularly and work on presentation skills. As automated proof generation advances, oral explanation and the ability to convey understanding may become more important ways of demonstrating mathematical contribution.
• Think about the value you want to create. Consider how your work might contribute to applications, education and problem-solving, or the enjoyment and understanding of mathematics.
• Tomorrow, take a long walk with a fellow student. Discuss why each of you entered mathematics, what you value about it, and how you would like to spend the next few years.

The speaker’s central reason for calling this a particularly good time to enter mathematics is that the field is actively reconsidering its norms. Newcomers may have more opportunity than usual to influence what mathematical research, teaching, and academic recognition become.

Speakers and sources featured

• Narrator: The presenter of 3Blue1Brown; the subtitles do not name the narrator explicitly.
• Konstantin Kogler: Featured in an interview excerpt discussing successful AI-assisted mathematical work.
• Other people, organizations, and examples mentioned: Paul Erdős; René Descartes; Évariste Galois; Alexander Grothendieck; Martin Gardner; Brady Haran; Numberphile; DeepMind; OpenAI; the Simons Foundation; and the U.S. National Science Foundation (NSF).