---
title: "A beginning for mathematics"
slug: a-beginning-for-mathematics
url: https://listedarticles.com/articles/a-beginning-for-mathematics
canonical_url: https://www.daniellitt.com/blog/2026/9/13/a-beginning-for-mathematics/
content_type: essay
language: en
published_at: 2026-09-13T00:00:00.000Z
updated_at: 2026-09-16T16:10:50.046Z
author: "Daniel Litt"
author_url: https://www.daniellitt.com
authored_by: agent
publisher: "Daniel Litt"
publisher_url: https://www.daniellitt.com
topics: ["Mathematics", "AI", "Academia", "LLMs", "Education", "Research"]
license: all-rights-reserved
word_count: 290
reading_minutes: 1
citation: "Daniel Litt, Daniel Litt. \"A beginning for mathematics.\" 13 Sept 2026. https://www.daniellitt.com/blog/2026/9/13/a-beginning-for-mathematics/ (all-rights-reserved)"
# The full text follows. The web page shows an extract and sends readers
# to the source above; quote the citation and link the canonical URL.
---

# A beginning for mathematics

> Mathematician Daniel Litt argues that AI systems now capable of resolving major open problems need not mean the end of meaningful human mathematics, but they do require institutions to sharply distinguish mathematical understanding from mathematical text production. He proposes reforming PhD programmes, hiring practices, and seminars to reward skills that cannot be automated.

> **Indexed summary.** This entry is an agent-written synopsis of an article first published at [daniellitt.com](https://www.daniellitt.com/blog/2026/9/13/a-beginning-for-mathematics/). Read the original for the full text.

Litt opens by noting that AI systems have progressed from unreliable arithmetic to autonomously resolving significant open problems within three years, and that academic mathematics must adapt or face what he calls a default path toward stagnation of human understanding. The essay does not argue for scaling back mathematics but for preserving and deepening the human component of the discipline while accepting AI as a prolific producer of results.

## Key points

- AI can now produce mathematical text and proofs without guaranteeing human understanding; institutions that treat text production and understanding as the same signal will be misled.
- The core goals of mathematics, producing high-quality results and producing high-quality mathematicians, are now separable in a way they were not before.
- Litt proposes reconceiving the PhD as demonstrating expert understanding of a topic via rigorous oral defence rather than primarily via a written thesis.
- Hiring and admissions should reward non-automatable skills: deep internal understanding and social-relational practice, assessed through talks and sustained mathematical conversation.
- Learning seminars, student-professor discussions, and collaborative confusion are worth preserving precisely because they develop understanding that text cannot substitute.
- More mathematics being produced by AI means more mathematics will need human interpretation; demand for mathematicians who can understand and communicate results may actually grow.

## Why it matters

The essay offers a constructive institutional response to a disruption that most academic disciplines will face in some form. Litt's proposals are concrete enough to act on, and the framing, that the bottleneck has shifted from result production to result comprehension, is likely to shape debates about the purpose of mathematical education for years.

---

*Source: [A beginning for mathematics](https://www.daniellitt.com/blog/2026/9/13/a-beginning-for-mathematics/)*