Anima Anandkumar is Accelerating Scientific Discovery with AI

Scientific progress is often limited not by a lack of new ideas, but by the cost and complexity of testing them. New solutions are needed to make that testing easier—and researchers like Anima Anandkumar are leading the way. She has conducted cutting-edge research across academia and industry for over a decade, pioneering new AI algorithms that simulate physical systems with unprecedented speed and accuracy—in some cases, over a million times faster than traditional methods. By empowering AI to model these systems, her research has unlocked advances across science and engineering, from high-resolution weather forecasting to designing novel medical devices.

[time-brightcove not-tgx=”true”]

“What fascinates me is how to bridge the gap between theory and practice, because I started at a time when deep learning wasn’t there—you had to start from first-principles design methods,” says Anandkumar, who explains that her approach to designing algorithms builds on fundamental principles found in maths and physics. Anandkumar is the Bren Professor of computing and mathematical sciences at Caltech, where she leads the Anima AI + Science Lab. She has also worked as a principal scientist at Amazon Web Services, designing machine learning-based solutions for Amazon cloud and a senior director of AI research at Nvidia. Informed by other scientific domains, particularly physics, she says her focus has always been on making algorithms “more principled, hardware efficient, and robust.”

Starting from this first-principles approach, Anandkumar and her collaborators developed “neural operators”: a kind of universal AI framework that can learn to simulate physical processes across multiple scales, from molecular interactions to climate patterns. Unlike large language models such as ChatGPT, AI models built with this framework can incorporate the laws of physics to test the plausibility of their predictions. And unlike traditional methods of simulating physical processes, which require immense computational resources to perform millions of calculations from scratch for each new prediction, these models are able to “learn shortcuts” from the data they’re trained on, Anandkumar explains—allowing them to simulate processes with equal or greater accuracy than methods that rely on raw computation, but at a much faster pace. Models designed in this way are particularly powerful because they “have the flexibility to learn the underlying continuous phenomena,” Anandkumar says.

In 2022, Anandkumar—in collaboration with an interdisciplinary team from Nvidia, Caltech, and other academic institutions—built a fully AI-driven open-source weather model, FourCastNet, using neural operators. It proved to be tens of thousands of times faster than the best “numerical weather prediction” models, while often also improving their accuracy. In less than two seconds, the model can produce a week-long forecast for a range of variables, such as wind speed and precipitation—what once required a supercomputer and several hours can now be done with far less hardware. It is available online via the European Centre for Medium-Range Weather Forecasts, and has inspired the adoption of similar weather models across the globe, despite initial skepticism from the climate modeling community. “This is already helping with extreme weather forecasts,” says Anandkumar, citing the model’s ability to accurately predict the path of Hurricane Beryl in June 2024, before conventional methods. 

Elsewhere, gains have been even more dramatic. In 2024, Anandkumar’s team worked with the U.K. Atomic Energy Agency to simulate the behavior of plasma in nuclear fusion reactors over a million times faster than prior techniques. This speed allows scientists to predict and prevent plasma disruptions—dangerous events where the super-heated plasma becomes unstable, which can damage the reactor if not caught early—before they occur, allowing technicians to preemptively take corrective action.

Anandkumar’s neural operators have proved useful not just for prediction, but also for design. The most common healthcare-related infections in the U.S. are catheter-associated urinary tract infections, which affect over a million Americans annually. In 2023, she and a team of Caltech researchers used their AI to prototype a catheter that reduced bacterial contamination one hundred-fold. They took a new approach: the model simulated fluid flow to identify where in the tube to place tiny grooves that prevent bacteria from swimming upstream to the patient’s body. The underlying AI framework can identify and test the feasibility of all sorts of designs, from drones to anti-cancer drugs.

Anandkumar’s work lights a path toward a future where AI and science reinforce one another: where scientific knowledge is deeply integrated with an AI’s understanding of the physical world, enhancing its capabilities; and where AI systems can generate and test new ideas. “Many labs, including us, are building towards this,” she says. “There’s so many discoveries that are happening as we speak.”

This profile is published as a part of TIME’s TIME100 Impact Awards initiative, which recognizes leaders from across the world who are driving change in their communities. The next TIME100 Impact Awards ceremony will be held on Feb. 10 in Dubai.

Source link