Brian Zhao

@atlasfellow 23 | IMO Bronze 21 23

[Brian Zhao @Brian_bfz]

You can read more about Mathathon in this @QuantaMagazine article by @kkakaes: https://t.co/QMASdcWxWJ

[Sathvik Redrouthu @_sathvikr]

We redesigned the Mathathon Challenge.

Introducing Mathathon 2.0.

Old problems, new proofs.

[Caltech Mathathon @math_hackathon]

13 days until the Mathathon! Teams are getting ready to spend 40 hours uncovering new secrets in old theories.

Here’s a sneak peak at the Definability of the Turing Jump:

In a 1999 paper, Slaman and Shore proved that the Turing jump, an operation that maps between decision problems of increasing difficulty, is definable in the first order theory of the Turing degrees. However, their work does not provide a simple definition of this jump. Teams will search for conditions in which an explicit definition is (or is not) attainable.

Want to learn more? Check out Slaman and Shore’s paper (https://t.co/nNLJNTTVw8). Application for judges is open at https://t.co/TWX0ss1yFO!

[Caltech Mathathon @math_hackathon]

15 days until the Mathathon! Teams are getting ready to spend 40 hours uncovering new secrets in old theories.

Here’s a sneak peak at the Robertson-Seymour Graph Minor Theorem:

This foundational result in modern graph theory proves that any family of graphs which is closed under taking minors can be characterized by a finite set of forbidden minors. Aspiring teams will work to simplify and shorten the elaborate proof of this result to make the theory more accessible and illuminate useful new arguments.

Want to learn more? Check out any of Robsertson and Seymour’s papers in the “Graph Minors’’ series (https://t.co/0I978qmxoO). Applications for judges are still open at https://t.co/TWX0ss1yFO!

[Caltech Mathathon @math_hackathon]

16 days until the Mathathon! Teams are getting ready to spend 40 hours uncovering new secrets in old theories.

Here’s a sneak peak at Iskovskikh’s classification of 3-folds:

Iskovskikh provides a classification of special Fano varieties, building off of a series of papers written by Gino Fano in the first half of the 20th century. Iskovskikh’s work has had significant impacts in the field of algebraic geometry, such as contributing to the success of Japanese mathematician Shigefumi Mori’s minimal model program. Teams will study Iskovskikh’s original classification and its broader implications, using current techniques in algebraic geometry to modernize, generalize, simplify, and contextualize this work.

Want to learn more? Check out Iskovskikh’s original paper (https://t.co/TgPCX91uT0) and find more problems at https://t.co/TWX0ss1yFO!

[Brian Zhao @Brian_bfz]

@SAIRfoundation @Caltech @xtxmarkets 🫡

[Brian Zhao @Brian_bfz]

@_sathvikr 👏

[Brian Zhao @Brian_bfz]

Applications are open for judges and speakers! Shape math how you’d like it to be shaped.

[Sathvik Redrouthu @_sathvikr]

We redesigned the Mathathon Challenge.

Introducing Mathathon 2.0.

Old problems, new proofs.

[Brian Zhao @Brian_bfz]

Honored to co-organize this event with SAIR and Terry.
Our goal is to bring together a diverse group of mathematicians to jointly set standards for responsible AI use.

[SAIR @SAIRfoundation]

The @Caltech community and SAIR are redesigning and co-organizing Mathathon: Old Problems, New Proofs.

Explore alternative proofs and clearer explanations, using AI to deepen human understanding.

More on our co-founder Terence Tao’s blog:
https://t.co/1cw7Zk2Jra

[Daniel Litt @littmath]

I've written an essay on how I think the mathematics profession should adapt to highly capable AI systems. It's hosted here on "Proofs and Prompts": https://t.co/sFTE8G4POQ
though you should also feel free to complain/comment on my website here: https://t.co/oyKN4xj59m

[Brian Zhao @Brian_bfz]

@ZGlaser97003 We'll set the red lines based on what we think the community consensus is. Happy to hear your thoughts as well!

[Brian Zhao @Brian_bfz]

I'm humbled by the amount of attention Mathathon has received and grateful to everyone who has offered feedback. It's through open and transparent discussions like this that we can progress as a field.

[Sathvik Redrouthu @_sathvikr]

We are aware of an open letter circulating online regarding the Mathathon.

We wrote a response here: https://t.co/Y6ywtYE3y4

We understand that it may not be sufficient to address all concerns, but we see this as a concrete first step. We are collaborating with the math community while actively exploring ways to redesign the event (moving away from the hackathon framing, adding a 6 month research period, etc).

We still believe in the value of an event that lets our participants explore, collaborate, and learn, with AI as a responsible aid. We welcome all feedback.

[Brian Zhao @Brian_bfz]

@ZGlaser97003 True, it is important to set red lines: academic misconduct, military use, etc. If someone repeatedly crosses these lines, they are a bad actor that cannot be reasoned with.
I don't know what these red lines are. We'll talk to a lot of mathematicians in the coming day.

[Alvan Arulandu @AlvanArulandu]

Over the last few years, I have had wonderful experiences in directed reading programs, NSF REUs, workshops, and conferences. It is important to explore ways these structures can adapt to best support early-career researchers in light of AI, without shunning it.

[Elliot Glazer @ElliotGlazer]

1. "The creation of slop mathematics and the exacerbation of a market-driven arms race. Undergrad participants may lack the expertise to carefully verify the results they produce. Participants are not required by the rules of the competition to communicate their results on ArXiv after the event is over. It will again fall to the mathematical community to carefully verify hastily-produced AI-generated results."

No one is forcing mathematicians to do anything here. Math academia has to decide for itself how to adapt verification labor norms in the era of cheaply produced mathematical outputs. And isn't the ArXiv *non-requirement* better for keeping it safe from unverified outputs?

2. "Intrusion of corporate interest. Events like these provide further financial and social leverage for AI corporations over mathematics. The goals of these corporations are not necessarily aligned with the aims of scientific research. As an independent scientific community, it is incumbent upon us to engage critically with AI companies and their tools. It is doubtful that such critical engagement will be incubated at the Mathathon."

Sorry but math academia does not have a monopoly on mathematical practice. As labs become increasingly mathematically productive, they do, in fact, an interest in engaging with aspiring mathematicians. Trying to prevent them from interacting with undergrads is foolhardy.

3. "Conditions unlikely to foster mathematical understanding. As most mathematical researchers would agree, forty hours is not enough time to deeply understand, solve, and communicate a problem. The industrialized production of AI-generated mathematical content on extremely shortened timescales does not add value to our community. Such activities in fact carry broadly negative impacts."

One does not become a research mathematician in 40 hours, no matter how they spend it. This would be a brief exposure to an essential modern tool to supplement their 4-year math major curriculum.

4. "Damaging misinformation about research mathematics. This event is likely to feed into detrimental narratives about our research community. As a case in point, the website of the Mathathon explicitly asks “What is the role of a mathematician when AI can solve conjectures faster?” The claim that AI can solve conjectures faster than human mathematicians appears both motivated and empirically dubious."

Grasping at straws. Someone in the future might make a bad narrative because of this 3-day event? That's an argument?

All the other comments are paternalist bullshit convinced they know better than undergrad math majors how they should be engaging with the changing landscape and deciding for them that they will be harmed by $20k in AI credits to explore what has quickly become an essential tool for math research.

Y'all should be ashamed of yourself for signing this bullshit.

[Caiman Moreno-Earle @Mav_Gators]

This has been a very long journey, and while specific policies and structures may change, our mission remains the same.

To help all of us mathematicians find a responsible path to a post-ai future, while preserving the discipline we love and have spent our lives exploring.

[Sathvik Redrouthu @_sathvikr]

A key deliverable of the Mathathon Challenge is the establishment of strong, responsible policy that lasts beyond the hackathon.

We invite a speaker panel of pro-AI and anti-AI mathematicians to honestly voice their opinions, and have been collaborating with both camps to design the best rules.

[Brian Zhao @Brian_bfz]

@Exomnium By EOY, idt AI will solve more problems in a world with mathathon than the counterfactual. It only changes the number of problems solved & verified in an academic vs hobbyist setting.

[Brian Zhao @Brian_bfz]

@hsalis What would the bio version entail? And why wait a few years?

[Brian Zhao @Brian_bfz]

@aixparty @oprydai yep we cover international flights.
you need to get your own visa tho.