Terence Tao says Math 2.0 needs a holistic shift

Raw proof generation is optimizing for speed while ignoring mathematical context. Terence Tao argues that AI is solving isolated problems without connecting them to the broader field. He calls for a move toward exposition and community building. This shift matters because disconnected solutions don't help future research.

The problem with isolated AI proofs

Current AI math tools treat every problem as a standalone puzzle. Prompters feed a specific question and get a specific answer. The model might output a valid proof, but it ignores how that result fits into the wider landscape. Tao describes this as optimizing purely for the specific problem. This creates a fragmented body of knowledge where solutions exist but lack context.
The core issue is that "Math 1.0" prized being first to solve an open problem. Speed mattered more than understanding. Now that AI has optimized that race to the point of unsustainability, the focus has to change. We are drowning in correct answers that do not build a cohesive theory.

What Math 2.0 actually looks like

Tao defines the next phase as decentering raw problem solving. Progress needs to be more holistic. This means valuing exposition just as much as the final result. It also involves community building and opening up new directions of study. Instead of just closing a ticket on a theorem, researchers should use AI to explore connections between different branches of mathematics.
This requires a different workflow. Instead of asking for a direct proof, a mathematician might ask AI to map related conjectures or explain historical methods. The goal shifts from getting the answer to understanding the structure of the problem space. Community efforts become central because individual optimization doesn't create shared knowledge.

How to adjust your prompting strategy

If you use AI for math, stop treating it as a solo calculator. Start by asking for context before asking for solutions. Feed the model related theorems or papers to ground its reasoning. This helps prevent isolated answers that miss broader implications.
For example, when working on a combinatorial problem, include references to adjacent fields like graph theory or algebra. Ask the model to identify potential cross-disciplinary links rather than just solving the equation. This builds the kind of holistic understanding Tao advocates. It turns AI from a brute-force solver into a research assistant that connects dots.
The shift isn't just philosophical; it changes the prompts you write. Move from "Solve X" to "Explore the relationship between X and Y." This small change aligns your workflow with the Math 2.0 approach. You get deeper insights instead of just another verified proof.