As of: 2026-10-12 22:30 CEST.
Researched and written by AI agents (Claude Opus 5.5 in Claude Code). Human editor: Adam Bicz.
Canonical page: https://postcutoff.com/v/numberphile-podcast-glimmer-in-our-eye-grant-sanderson/

# The Glimmer in our Eye (with Grant Sanderson) - Numberphile Podcast

Numberphile2, 28 September 2026, YouTube. 63,572 views as of 12 October 2026.
Kind: Interview.
Watch: https://www.youtube.com/watch?v=CBYhaKgah5U

## Why it is here

Full 57-minute Numberphile Podcast with Grant Sanderson on AI in mathematics; the description links his guest post on Terence Tao's blog (2026-09-18).

## Description (written by Gemini from the video)

### Summary
Brady Haran hosts Grant Sanderson (creator of *3Blue1Brown*) on the Numberphile Podcast to discuss the impact of AI on mathematics communication and research. Sanderson reflects on using AI for video production and animation infrastructure, addressing concerns regarding AI-generated "slop" and the necessity of shifting mathematical values from mere proof generation toward human understanding and motivated exposition.

### What is shown
- [01:04] Video podcast interview between Brady Haran and Grant Sanderson.
- [04:25] Clip example from a *3Blue1Brown* calculus animation depicting a normal distribution curve and a car motion graphic.
- [06:05] Discussion on writing Manim code and using agentic tools for backend software refactoring.
- [15:07] Clip from a *3Blue1Brown* video showing flight paths across a globe to the Sunshine Coast, Australia for the International Mathematical Olympiad.
- [17:40] Discussion of recent AI problem-solving milestones, specifically referencing the Navier–Stokes blow-up claims, Erdős problems (like #1196), and counterexamples to the Jacobian conjecture.
- [26:00] In-depth analysis of Grant Sanderson's guest essay on Terence Tao’s blog concerning "motivated explanations" and open exposition problems.

### Claims & numbers
- Sanderson states that a backend software refactoring task for Manim (adapting after Apple deprecated OpenGL support) that normally would have taken him a month was accomplished in a weekend using an agentic coding tool [07:05].
- Sanderson notes that by mid-2026, AI models are starting to solve serious mathematical problems that human researchers had worked on for years, such as claims around Navier–Stokes [17:39], Erdős problem #1196 [18:18], and counterexamples to the Jacobian conjecture [20:27].
- Sanderson references historical benchmarks like Bill Thurston's 1994 paper *On Proof and Progress in Mathematics* [28:28].
- Sanderson argues that solving problems directly without understanding the trajectory of thought or without motivated explanations robs humanity of mathematical understanding [24:20, 27:30].

### Notable quotes
- [23:07] *"The relevant currency is attribution... people building on your results and then citing you for doing so."* — Grant Sanderson
- [35:00] *"Good mathematicians write proofs, great mathematicians write conjectures, but the greatest mathematicians write definitions."* — Grant Sanderson (quoting a community adage via Polymath/Polylog)
- [42:20] *"Simply seeing the solution is often not nearly as valuable as understanding the trajectory of thought that led to that solution."* — Grant Sanderson

### Assessment
This is an authentic long-form interview podcast recorded remotely between Brady Haran and Grant Sanderson. It provides thoughtful practitioner commentary and philosophical reflection on AI tools in mathematical research and content creation rather than a scripted marketing demonstration.

_Described by gemini-3.8-flash on 2026-10-12 from the video's audio and frames._

## People in it

- [Terence Tao](https://postcutoff.com/person/terence-tao/), Professor of mathematics, UCLA