Math, Inc. has just released an exciting new artificial intelligence tool. Meet Gauss, a new kind of reasoning agent that’s changing the game on how we verify mathematical proofs. CEO and co-founder Jesse Han revealed on February 23 that Gauss has achieved a significant milestone in autoformalization and AI–human collaboration. This novel approach automates the formalization of complex mathematical proofs. It further enhances the partnership between human mathematicians and AI.
Gauss aims to supplement and eventually integrate more traditional natural language reasoning with fully formalized reasoning. This new capability allows it to perform systematic literature searches and run reproducibility verification tools. It is able to produce code in Lean, a burgeoning programming language for encoding formal proofs. Gauss’s capabilities extend well beyond basic automation. They herald in a new era where human intelligence joins with machine efficiency in a powerful new partnership.
A New Era of Proof Formalization
Gauss has demonstrated its formidable capabilities by successfully autoformalizing Viazovska’s 24-dimensional sphere packing proof, which consists of over 200,000 lines of code, in just two weeks. This achievement is particularly impressive as it highlights the model’s incredible ability to work through difficult areas of mathematics. It can produce all-encompassing, provably correct solutions orders of magnitude faster than any human mathematician.
Not only the 24-dimensional proof, Gauss also formalized the 8-dimensional sphere packing proof in a mere five days. Importantly, it tracked down and even fixed a typo in the original published paper about this proof. Han commented on the complexity of the 24-dimensional sphere packing proof, noting, “It was actually significantly more involved than the 8-dimensional case, because there was a lot of missing background material that had to be brought online surrounding many of the properties of the Leech lattice, in particular its uniqueness.”
The success of Gauss is due to the fact that it makes use of very powerful mathematical functions called (quasi) modular forms. We can only imagine the ways these functions helped Gauss to comprehend and navigate the details of the Leech lattice, all while formalizing the proof of 24 dimensions. This is a remarkable breakthrough for the state of mathematical research as well as the field of AI.
Collaborating with Human Mathematicians
Gauss’s final development was informed by prior research led by academics like Hariharan and colleagues. After a partial breakthrough in mid-January, Gauss was ordered to return his attention mainly on preparing his work for publication formalizing sphere packing. While developing the concept in this phase, the team worked in conjunction with mathematicians. Their efforts focused on formalizing prior but verified proofs, a great example of how AI can supplement human work rather than replace it.
The collaboration yielded impressive results, with Hariharan stating, “They told us that they had finished 30 ‘sorrys’, which meant that they proved 30 intermediate facts that we wanted proved.” This underscores the model’s power to address multidimensional challenges and offer substantial advancements to real-world research still in progress.
Liam Fowl noted the rapid advancements signaled by Gauss’s performance, stating, “These new results seem very, very impressive, and definitely signal some rapid progress in this direction.” These endorsements are indicative of the enthusiasm in the mathematical community about the possibilities that AI offers for accelerating future discoveries.
The Future of Mathematics and AI
Gauss doesn’t just automate the drudgery, it fundamentally changes the way mathematicians work. This tool marks a new and collaborative chapter in their problem-solving. As Han articulated, “I think the end result of technology like this will be to free mathematicians to do what they do best, which is to dream of new mathematical worlds.” This expression comes close to describing the spirit of Gauss’s role—it is meant to enhance human creativity and intellectual exploration in mathematics.
This technology is changing quickly. Perhaps more importantly, it will soon begin driving research and proof verification processes. Han emphasized the transformative potential of Gauss by explaining its multifaceted capabilities: “It’s able to conduct literature searches, call up tools, and use a computer to write down Lean code, take notes, spin up verification tooling, run the Lean compiler, etc.”
The effects of Gauss’s miracles go beyond the direct day-to-day rewards to mathematicians. More fundamentally, they ask what the future role of AI should—and must—be in relation to human expertise across all professions and industries. As Han remarked on the evolution of programming, “A programmer used to be someone who punched holes into cards, but then the act of programming became separated from whatever material substrate was used for recording programs.” This comparison illustrates the way in which emerging AI technology has the potential to dramatically shift these same roles in both academic and industry spaces.