For over a decade, MIT Associate Professor Rafael Gómez-Bombarelli has been at the forefront of using artificial intelligence to innovate new materials. As AI technology continues to evolve, so too do his ambitions. Now a newly tenured professor in materials science and engineering, Gómez-Bombarelli believes AI is set to revolutionize science in unprecedented ways. His work at MIT and beyond is dedicated to accelerating this future.
“We’re at a second inflection point,” Gómez-Bombarelli asserts.
The first one was around 2015 with the first wave of representation learning, generative AI, and high-throughput data in some areas of science. Those are some of the techniques I first brought into my lab at MIT. Now I think we’re at a second inflection point, mixing language and merging multiple modalities into general scientific intelligence.
Gómez-Bombarelli’s research blends physics-based simulations with methodologies like machine learning and generative AI to unearth new materials with promising real-world applications. His work has already led to advancements in materials for batteries, catalysts, plastics, and organic light-emitting diodes (OLEDs). Additionally, he has co-founded several companies and served on scientific advisory boards for startups applying AI to fields such as drug discovery and robotics. His latest venture, Lila Sciences, aims to construct a scientific superintelligence platform for the life sciences, chemical, and materials science industries.
From Experiments to Simulations
Gómez-Bombarelli’s journey into the realm of AI and materials science began in Spain, where he developed an early interest in the physical sciences. In 2001, he won a Chemistry Olympics competition, setting him on an academic path in chemistry at the University of Salamanca. There, he pursued both undergraduate and PhD studies, focusing on the function of DNA-damaging chemicals.
During his PhD, Gómez-Bombarelli transitioned from experimental work to simulations and computer science. “I started simulating the same chemical reactions I was measuring in the lab,” he recalls.
I like the way programming organizes your brain; it felt like a natural way to organize one’s thinking.
His fascination with simulations led him to a postdoctoral position in Scotland, studying quantum effects in biology, before joining Alán Aspuru-Guzik at Harvard University in 2014.
By 2016, Gómez-Bombarelli was among the pioneers using generative AI for chemistry, applying neural networks to understand molecular structures. His efforts to automate molecular simulations resulted in hundreds of thousands of calculations across materials, identifying numerous promising candidates for further testing.
Building a Computational Future
In 2018, after a stint in industry, Gómez-Bombarelli was encouraged to apply for a faculty position at MIT. Despite initial reluctance, he was drawn to MIT’s dynamic and collaborative environment. “Everything I had been doing as a postdoc and at the company was going to be a subset of what I could do at MIT,” he says.
Today, his lab focuses on the interplay between the composition, structure, and reactivity of atoms and their impact on material performance. They employ high-throughput simulations to create new materials and develop tools that integrate deep learning with physics-based modeling.
Physics-based simulations make data and AI algorithms get better the more data you give them. There are all sorts of virtuous cycles between AI and simulations.
Gómez-Bombarelli’s group is entirely computational, allowing them a wide breadth of research without the constraints of physical experimentation. They collaborate closely with experimentalists and industries to ensure practical applications for their discoveries.
The Future of AI in Science
The rise of AI has seen the field mature significantly, with major companies like Meta, Microsoft, and Google’s DeepMind conducting physics-based simulations reminiscent of Gómez-Bombarelli’s early work. In November, the U.S. Department of Energy launched the Genesis Mission to leverage AI for scientific discovery and national security.
“AI for simulations has gone from something that maybe could work to a consensus scientific view,” Gómez-Bombarelli notes.
Humans think in natural language, we write papers in natural language, and it turns out these large language models that have mastered natural language have opened up the ability to accelerate science.
Reflecting on his time at MIT, Gómez-Bombarelli appreciates the non-competitive, positive-sum environment that fosters innovation. His research group, comprising around 25 graduate students and postdocs, embodies this ethos. “We’ve naturally grown into a really diverse group, with a diverse set of mentalities,” he says.
Everyone has their own career aspirations and strengths and weaknesses. Figuring out how to help people be the best versions of themselves is fun.
As AI continues to advance, Gómez-Bombarelli remains committed to pushing the boundaries of scientific exploration, ensuring that AI’s role in science is one of progress and positive impact.
