Interview: The Hidden Math Behind Everything with Jordan Ellenberg | Particles of Thought

Interview: The Hidden Math Behind Everything with Jordan Ellenberg | Particles of Thought

by Jordan Ellenberg

Formal & Physical Sciences Mathematics PBMathematicsPBBPhilosophy of mathematics
🎙 Jordan Ellenberg 👥 1.5M 📅 August 27, 2026 ⏱ 76 min 👁 84 📄 expert opinion 🧭 2026-08-27
Available in: English (current) Français

Keywords

mathématiquesIAgéométrieprobabilitésinterview

Summary

In this interview, mathematician Jordan Ellenberg discusses the pervasive nature of mathematics, arguing that it is not just about numbers but about the underlying structure of everything. He explains how AI models, particularly large language models, are trained through a process of trial and error known as gradient descent, which is fundamentally a mathematical optimization. Ellenberg also touches on the concept of embeddings in high-dimensional spaces, which represent words as points in a geometric space, and contrasts this with human language evolution, which he argues is not easily predictable through differential equations. The conversation explores the random walk concept, illustrated by the journey of photons through the sun, and its historical significance in physics and information theory, referencing Einstein and Claude Shannon. Ellenberg shares his perspective on AI as a tool for mathematical discovery, particularly in generating interesting examples, but emphasizes that it remains a tool guided by human direction. The discussion also briefly touches on the status of string theory, noting its lack of experimental confirmation. Throughout, the interview highlights the deep connections between mathematics, physics, and artificial intelligence, and the importance of uncertainty and probability in understanding the universe.

191 words

Critical Evaluation

Value of the Information & Strength of the Argument

The value of the information lies in the expert insights provided by Jordan Ellenberg, who demystifies complex mathematical concepts and connects them to everyday phenomena and cutting-edge technology. The argumentation is solid, as Ellenberg supports his claims with clear explanations and references to established mathematical principles and historical examples. For instance, he explains gradient descent in a simple, intuitive manner, and uses the random walk to illustrate both physical processes and the foundations of information theory. The discussion is engaging and thought-provoking, encouraging a broader appreciation of mathematics. However, some arguments are presented as opinions (e.g., on the limitations of predicting language evolution) without deep empirical backing, but this is appropriate for an interview format.

Scientific Rigor, Source Quality, Title Accuracy

The scientific rigor is high, given the credentials of the participants. Ellenberg is a professor of mathematics and a bestselling author, and Oluseyi is an astrophysicist with extensive research experience. The sources cited are primarily conceptual, referencing the work of Einstein, Shannon, and the general principles of machine learning. The title accurately reflects the content, which is an interview about the hidden math behind various aspects of the world. The description provides additional context and links to the podcast platform and NOVA’s newsletter, which are relevant for further exploration. The discussion is well-structured and stays on topic, though it occasionally veers into personal anecdotes, which adds to its accessibility.

239 words

Title / Content Match

The title accurately reflects the content: an interview exploring the pervasive role of mathematics in various domains, from AI to physics and language.

Quality & Reliability

8/10

The discussion is led by a renowned mathematician (Jordan Ellenberg) and an astrophysicist (Hakeem Oluseyi), providing expert perspectives. The content is largely conceptual and explanatory, with references to established mathematical concepts (e.g., gradient descent, random walks, information theory) and historical figures (Einstein, Shannon). While not a formal academic review, the expertise of the speakers and the alignment with established scientific knowledge support a high reliability score.

Key Moments

Markers derived by PSI from the transcript: the creator did not define chapters.

Cited Sources

Concurring Sources

  • Gradient Descent - Wikipedia — Provides a formal definition and explanation of gradient descent, aligning with Ellenberg's description.
  • Random Walk - Wikipedia — Offers a comprehensive overview of random walks, supporting the discussion on photons and language.

Dissenting Sources

External References

Contribution & Novelties

The interview provides an accessible yet expert perspective on the ubiquity of mathematics, particularly in the context of AI and physics. It offers a clear explanation of gradient descent and embeddings, making these concepts understandable to a general audience. The discussion also highlights the historical roots of modern AI in information theory and the random walk, connecting disparate fields. The novelty lies in the conversational format that bridges mathematics, physics, and AI, offering insights that are both educational and thought-provoking.

Pour aller plus loin :

  • Gradient descent — Core optimization algorithm discussed in the context of AI training.
  • Random walk — Mathematical concept illustrated with photons and language models.
  • Information theory — Foundational to understanding Shannon’s work and modern AI.
  • Embedding — Representation of words in high-dimensional spaces, as discussed in the interview.

133 words

Radar Profile

The radar profile shows high scores in quality and reliability, reflecting the expert nature of the discussion. The quantity of information is moderate, as the interview is more conceptual than data-dense. The technical level is high, but accessible, making it valuable for an informed audience.

Reliability 8/10

💬 No comments were provided for analysis.