An Open Commitment to Mathematics and Science

Today SAIR names Susquehanna International Group and XTX Markets as the first supporters of its open research program.

On September 11th, twenty-five Fields Medalists, including SAIR co-founder Terence Tao, published a declaration on the state of AI in mathematics. Its central point is simple. Solving a famous problem has always been a sign of new understanding. The value was never the answer alone. It was the ideas, the methods, the careful write-ups, and the students and colleagues who carried them forward. As AI tools accelerate the production of mathematical results, openness becomes more important: new results should come with methods that can be inspected, verified, understood, and built upon.

The same holds beyond mathematics. Physics, chemistry, biology, and every other experimental science are now asking how AI should enter their work. Mathematics is where the question arrived first because its results can be fully verified, and its problems are the most clearly stated. It is the right place to establish the model.

SAIR was founded to build that model: open infrastructure for research in mathematics and the sciences that depend on it. These are the terms we operate under.

Problems are chosen by researchers. Working scientists and mathematicians design every SAIR competition at the research level, including open problems with no known answer.

The competition work takes place in public. Teams collaborate in an open forum, with the organizers answering questions in the threads. Anyone can read along.

Research outputs are released openly. Data, benchmarks, code, and verified results are released under open licenses, so the community can reproduce them, build on them, and cite the people who did the work.

Credit is preserved. Contributors are named. Methods are written up. Nothing is announced before it can be checked.

Humans stay in the loop. AI is a tool for accelerating understanding, and the researchers who use it remain responsible for what enters the record.

XTX Markets is supporting the three SAIR competitions launching this fall: the Lean Kernel Challenge, co-organized with the Lean FRO; the Andrews-Curtis Conjecture Challenge, co-organized with Caltech; and Stage 2 of the Inverse Galois Challenge, co-organized with the LMFDB. Susquehanna is supporting the Lean Kernel Challenge and will work with SAIR to co-create a future competition, with a broader commitment to supporting the program as it develops. Both firms have long-standing ties to the mathematical community and are helping make rigorous, open research available for others to verify, use, and build upon.

"We are incredibly thankful to have the support of XTX Markets and Susquehanna, forward-thinking collaborators and staunch believers in an open research ecosystem," said Terence Tao. "Their early backing lets problems come from researchers, work happens in the open, and results be released for everyone to build on. That is what an open commitment to mathematics and science looks like in practice."

"At Susquehanna, we see firsthand what mathematics and artificial intelligence can achieve together. As these capabilities evolve, it is increasingly important that new ideas, methods, and tools contribute to a rigorous and open scientific ecosystem. We are delighted to support SAIR as it develops new ways to enable ambitious, collaborative research in the AI era," said Rob Rhoades, co-head of Quantitative Research for Options Trading at Susquehanna.

Simon Coyle, Head of Philanthropy, XTX Markets said, "Artificial intelligence offers us a potential golden age for mathematical discovery. To make this happen, we will need a combination of AI tools, including frontier and open-source, and robust norms to promote high-quality mathematics. We are delighted to be partnering with SAIR on these open competitions, which we believe offer a valuable model for advancing this vision."

Researchers can register for the fall competitions at: competition.sair.foundation.

Organizations interested in supporting open research in mathematics and science can contact math@sair.foundation.