Terence Tao uses coding agents to bring mathematical visualizations back to life
A practical example of using AI coding assistance for sandboxed, deterministic mathematical applets.
Visual references
3 images
Small visual tools are a promising use case
Terence Tao's recent post describes modernizing old Java applets and creating new visualizations with coding agents. The examples include a stochastic process related to the zeta function and a playful model of celestial motion.
Why this use case is different
These applications are secondary visual aids: they run locally, produce deterministic output, and are not the authority for a theorem or a safety-critical decision. That makes it easier to tolerate a small bug while still requiring basic testing. The post is a useful counterweight to the idea that every AI-generated program should be trusted—or that every program must be hand-written to be educational.
A design pattern for MathsAI
Give learners a visual model, expose its assumptions, and label it as an aid rather than a proof. Let students change parameters, predict what will happen, and compare the animation with a calculation.