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

## Executive summary

The video analyzes the future of mathematics in the age of advanced AI, arguing that while machine capabilities (such as generating proofs for problems like the Navier-Stokes equations or solving IMO-level geometry) will fundamentally change the academic credit system, the core value of mathematics remains robust. The value shifts from the proof itself to 'theory building'—the ability to synthesize novel ideas—and the human skills of problem-solving, communication, and aesthetic joy. The speaker advises that the most valuable human contributions will be in theory building, oral presentation, and fostering community understanding.

## Key takeaways

- The Shift from Proof to Theory Building: The value of math is shifting away from the proof of a theorem (which is increasingly viewed as 'raw data' by AI) toward 'theory building'—the creation of broad conceptual frameworks (e.g., Galois's work on group theory or Grothendieck's algebraic geometry).
- The Role of Human Skills in the AI Era: The ability to solve problems, even if machines are better at generating proofs, remains a crucial skill (analogized to going to the gym). Furthermore, the most relevant human skills are now seen in oral presentation and communicating complex ideas to others.
- The Value of Community and Joy: The aesthetic joy and the communal nature of mathematics—the pleasure of learning and sharing ideas—are considered intrinsically valuable and are less affected by AI than the technical components.
- The Future Focus: Theory and Communication: If AI handles proof generation, the focus for human research must be on theory building and understanding how those generated proofs can be woven into a coherent, novel theory. The speaker suggests that the most effective measure of human potential will be the ability to give talks and communicate ideas.

## Technical details

- AI Capabilities in Math: AI models (LLMs) are capable of solving complex problems, such as IMO-level geometry questions and generating proofs for serious conjectures (e.g., Paul Erdős problems). The speaker notes that the pace of progress over the last four years has been unprecedented.
- Mathematical Value Stack: The value of math is composed of three parts: 1) Applications (e.g., imaginary numbers leading to quantum mechanics); 2) Problem-Solving Skills (transferable skills); and 3) Aesthetic Joy. The speaker argues that the first two are less affected by AI than the third.
- Key Mathematical Concepts: The discussion references specific fields and concepts, including: Cartesian coordinates (Descartes), Group Theory (Galois), Algebraic Geometry, Elliptic Curves, and the Navier-Stokes equations. The speaker emphasizes that the *theory* built around these concepts is more valuable than the proof of a single theorem.
- Communication and Attribution: A significant challenge is the attribution of machine-generated proofs, as models often repackage existing human-written ideas without proper credit. The speaker stresses the need for human effort to make machine output understandable.

## Practical implications

- Focus research efforts on 'theory building' rather than solely on generating proofs.
- Prioritize developing strong oral presentation and communication skills, as this is a highly relational and human-dependent skill.
- For students considering a career in math, the speaker suggests engaging in community discussion and reflecting on personal passion, rather than solely focusing on technical problem-solving.
- The most concrete advice is to 'go on a long walk' with peers to discuss what they genuinely love about math and how they want to contribute to the field's evolution.

## Topics

Artificial Intelligence, Mathematics, Theoretical Physics, Education, Academic Funding, Computer Science, 3blue1Brown Website, Proofs and Prompt, manim (Python library)

Source: https://www.youtube.com/watch?v=LKrwhjOI8Nc