A team of six UCLA faculty members spanning artificial intelligence, computer science and mathematics has received a $250,000 seed grant from the Laude Institute as part of its inaugural Moonshots initiative, which supports university-led efforts to solve some of humanity’s hardest, undiscovered problems.
The UCLA team was one of just eight selected from 125 proposals by more than 600 researchers across 47 institutions in the U.S. and Canada, as announced by the Laude Institute today. Led by Amit Sahai, a professor of computer science at the UCLA Samueli School of Engineering, other UCLA computer scientists include professor Raghu Meka, department chair Wei Wang, and associate professors Kai-Wei Chang and Nanyun (Violet) Peng. Terence Tao, a UCLA mathematics professor and director of special projects at the Institute for Pure and Applied Mathematics, is part of the team as well.
“I’m most excited by our goal of creating AI capable of wonder-based discovery in mathematics — an AI system that not only solves problems rigorously, but also asks what might be true and explores the boundary between possible and impossible,” Amit Sahai said.
Titled “Accelerating the Queen of Sciences,” the project seeks to develop an AI system for rigorous and creative mathematical thinking, ultimately serving as a research partner for scientists across every quantitative field. Rather than training AI to solve known problems, the team is focused on teaching AI to behave like a mathematician and ensure claims are automatically verified, turning ‘sounding right’ into ‘being right.’
“I’m most excited by our goal of creating AI capable of wonder-based discovery in mathematics — an AI system that not only solves problems rigorously, but also asks what might be true and explores the boundary between possible and impossible,” said Sahai, who holds the Symantec Term Chair in Computer Science.
Today’s leading AI models can solve competition-level math problems and generate fluent explanations, but they often struggle with sustained, multi-step reasoning and do not provide reliable proofs. They can also produce answers that seem convincing yet are incorrect.
To address these limitations, the UCLA team is developing an AI-powered system that learns to think and reason like a mathematician — working through entire textbooks, generating new problems, iterating on failed attempts and building deeper conceptual understanding.
The approach also features an exploration mode, where the model freely speculates about hidden patterns and open questions, paired with a verification layer that evaluates those ideas. By integrating formal proof systems, the researchers ensure that outputs adhere to strict standards of mathematical accuracy.
“The pace of progress in AI capabilities and formalization over the past few years has been truly remarkable,” said Tao, who holds the James and Carol Collins Chair and won a Fields Medal in 2006. “Previously impossible ‘moonshots,’ such as creating a rigorously verified, effective and genuinely creative explorer of mathematical results, now seem within reach. I’m excited to see what our UCLA team can achieve in this direction.”
“Previously impossible ‘moonshots,’ such as creating a rigorously verified, effective and genuinely creative explorer of mathematical results, now seem within reach,” Terence Tao said.
Over six months, the project will produce an open-source proof of concept, a benchmark suite for evaluating research-level mathematical reasoning, a lab proposal and a vision paper developed through a UCLA-hosted workshop with leading domain experts.
Following the seed phase, the eight winning teams will present their work in October to the selection committee, competing for $10 million to establish a multi-year Moonshot lab at the winning team’s institution.
Another UCLA research team, led by Wang and includes many of the same researchers, recently received a three-year, $5 million grant from the Defense Advanced Research Projects Agency to develop AI tools aimed at transforming how mathematical discoveries are made, formalized and verified.
Notable Laude Moonshots selection committee members include John Jumper, Nobel laureate in chemistry and a director at Google DeepMind; John Hennessy, Turing Award winner, former president of Stanford University and now chairman of Alphabet’s board; and Eric Horvitz, chief scientific officer at Microsoft. The selection committee is chaired by David Patterson, Turing Award winner and founding board chair of the Laude Institute.
“What excites me about this team is that they’re not asking AI to solve known problems faster; they’re asking it to wonder, conjecture and discover the way a mathematician does,” said Patterson, a UCLA alumnus with a B.A. in mathematics and an M.S. and Ph.D. in computer Science. “This world-class team is tackling the fundamental bottleneck of reasoning itself. If they succeed, the implications extend far beyond mathematics — this project could unlock progress across every hard science.”
Laude, a nonprofit co-founded in 2025 by Perplexity and Databricks co-founder Andy Konwinski, aims to bridge the gap between academic AI research and real-world impact. The organization’s Moonshots initiative focuses on accelerating scientific discovery, improving health care delivery, strengthening civic discourse and helping workers adapt for the AI age.