Unleashing the Power of AI and Simulations to Revolutionize Science
Imagine a future where scientific research is not just efficient but seamless, where the potential for discovery is limitless. This is the vision that drives MIT's Associate Professor Rafael Gómez-Bombarelli, a pioneer in using artificial intelligence to create new materials.
The Future of Science is Here
Gómez-Bombarelli's journey began over a decade ago, and his ambitions have grown alongside the expanding capabilities of AI. Now, with his recent tenure as a professor in materials science and engineering, he believes we stand at the brink of a scientific revolution, one powered by AI.
"We're at a pivotal moment," he asserts. "The first wave of AI-driven scientific advancements began around 2015 with representation learning and generative AI. Now, we're merging language and multiple modalities to create a general scientific intelligence. We have the tools to reason about language, material structures, and synthesis recipes like never before."
From Theory to Practice
Gómez-Bombarelli's research is a fusion of physics-based simulations and machine learning, resulting in the discovery of new materials with real-world applications. His work has led to innovations in batteries, catalysts, plastics, and organic light-emitting diodes (OLEDs). But his impact extends beyond academia; he has co-founded multiple companies and served on advisory boards, applying AI to drug discovery and robotics.
His latest venture, Lila Sciences, aims to build a superintelligence platform for the life sciences and materials industries. It's a bold step towards ensuring that the future of scientific research is more productive and efficient.
"AI for science is an incredibly exciting and ambitious application of AI," Gómez-Bombarelli explains. "Unlike other AI applications, science has the potential to bring a better future forward in time, with fewer downsides and ambiguities."
The Evolution of Scientific Inquiry
Gómez-Bombarelli's path to AI-driven science began in Spain, where he developed an early interest in the physical sciences. A winner of the Chemistry Olympics in 2001, he pursued an academic career in chemistry, studying at the University of Salamanca. It was during his PhD that he discovered the world of simulations and computer science, a revelation that changed his trajectory.
"My PhD started with experimental work, but I got hooked on simulations. I began simulating the same chemical reactions I was measuring in the lab. Programming organized my thoughts in a way that felt natural, and it offered fewer limitations than hands-on experiments or scientific instruments."
Post-PhD, Gómez-Bombarelli ventured to Scotland for a postdoctoral position, studying quantum effects in biology. Here, he met Alán Aspuru-Guzik, a chemistry professor at Harvard University, who would become a key collaborator.
"I was an early adopter of generative AI for chemistry and one of the first to use neural networks to understand molecules. It was the infancy of deep learning for science."
Gómez-Bombarelli also worked to automate molecular simulations, running hundreds of thousands of calculations and discovering promising materials for testing. This led to the founding of a materials computation company, which later focused on producing OLEDs.
"It was challenging but rewarding to create something tangible. Seeing Aspuru-Guzik's energy and creativity running a 40-person lab made me reconsider my career path. I realized I didn't have the same level of energy, and I wanted a different kind of impact."
The MIT Journey
In 2018, Aspuru-Guzik encouraged Gómez-Bombarelli to apply for a position at MIT's Department of Materials Science and Engineering. Despite initial trepidation, Gómez-Bombarelli's thoughts on faculty life changed after his startup experience and the encouragement of Aspuru-Guzik. He was drawn to MIT's collaborative spirit and the vast research possibilities.
"Everything I had done as a postdoc and at the company was just a fraction of what I could do at MIT. I could still make products, but my work expanded exponentially. My universe of work became a subset of this new, exciting universe of possibilities."
Nine years later, Gómez Bombarelli's lab at MIT focuses on understanding how the composition, structure, and reactivity of atoms impact material performance. He has used high-throughput simulations to create new materials and developed tools to merge deep learning with physics-based modeling.
"Physics-based simulations and AI algorithms improve with more data. There's a beautiful synergy between AI and simulations."
His research group, solely computational, collaborates closely with experimentalists, creating tools to help them navigate AI-generated ideas.
"We're blessed with the ability to work on a broad range of projects simultaneously. We love working with experimentalists and creating tools to support their work."
Gómez-Bombarelli also maintains a focus on the real-world applications of his research, working with companies and organizations like MIT's Industrial Liaison Program to understand industry needs and development challenges.
The Impact of AI on Science
As AI has gained prominence, Gómez-Bombarelli has witnessed its maturation. Companies like Meta, Microsoft, and DeepMind are now conducting physics-based simulations, similar to his work in 2016. The U.S. Department of Energy's Genesis Mission aims to accelerate scientific discovery and national security using AI.
"AI for simulations has evolved from a potential tool to a consensus scientific view. We're at a turning point. Humans think and communicate in natural language, and large language models have mastered this. We've seen that scaling works for simulations and language. Now, we'll see how it scales for science."
Gómez-Bombarelli is inspired by the collaborative spirit at MIT, a stark contrast to the competitive nature he often sees in other institutions. He strives to foster a positive-sum environment in his research group, comprised of graduate students and postdocs.
"Our group is incredibly diverse, with a range of mentalities, career aspirations, strengths, and weaknesses. It's a joy to help people reach their full potential. I've even become the one pushing people to apply for faculty positions after the deadline!"
As we stand on the cusp of a scientific revolution, Gómez-Bombarelli's work and vision are a testament to the power of AI and simulations to accelerate discovery and bring a better future forward.
