Adji Bousso Dieng is using Artificial Intelligence to design ‘smart’ materials with tiny holes called nanopores that can selectively absorb and release small chemicals such as carbon dioxide or certain pharmaceuticals. Being able to design such structures would make it possible to have greater control over the movement and delivery of small molecules than is currently possible. For carbon dioxide, materials with nanopores might make it possible to build large-scale carbon capture systems that take carbon dioxide out of the atmosphere and concentrate so that it can be used industrially. For pharmaceuticals, nanopores might be critical to creating systems that could deliver small doses of a drug over an extended period of time, vastly simplifying the management of many diseases.
Her work will focus on Hard Problems 1 (develop more capable and more general AI, that is safe and earns public trust) and 4 (having AI address one or more of humanity’s greatest challenges and opportunities, including in health and life sciences).
In 2021 Adji joined Princeton University as an Assistant Professor in Computer Science. She also has associated faculty appointments in the Department of Chemical and Biological Engineering, the Princeton Materials Institute, Princeton’s graduate program in Quantitative & Computational Biology, and Princeton’s High-Meadows Environmental Institute.
Her lab is named Vertaix. In addition to her work in the sciences, she has also established a non-profit organization The Africa I Know, that seeks to inspire young Africans to pursue studies and careers in Science, Technology, Engineering, Mathematics, Artificial Intelligence, and more.
Learn more about Adji Bousso Dieng:
What does the name Vertaix mean? Where does it come from?
I love that question!
Vertaix refers to the idea of a vertex in geometry, the point where things connect. The reason why I named my lab Vertaix is because the vision I have for the lab is for AI to be the point where all the sciences connect with each other. Our work on the Vendi Score, the first paper from Vertaix, is a good highlight of that vision. The Vendi Score [a new way of measuring the performance of machine learning systems] was designed by taking inspiration from and extending ideas from ecology and quantum mechanics, two very different areas of the natural sciences.
So the name Vertaix reflects that desire to connect a diverse set of sciences to develop new AI methods to advance science. The name also reflects the desire to cultivate a diverse lab, composed of people from different backgrounds, different walks of life, and different skill sets.
What are you working on now?
At Vertaix we are currently working on, among other things, developing methods that are rooted in diversity [defined here in a scientific sense as the range of data and outputs from models used in machine learning]. This is because diversity is, I believe, at the core of many challenges in machine learning and science. For example, generative modeling, reinforcement learning, active search, molecular simulations, and more. All of these areas are most effective at delivering good results for science applications when diversity is incorporated. So in order to fulfill the promise of machine learning for science, we ought to carefully look into how to effectively measure and embed diversity in machine learning methods, which is what we have pioneered at Vertaix with the Vendi Score.
Any recent publications funded by AI2050, or are on the horizon?
Yes, we put out two papers about diversity:
The Vendi Score: A Diversity Evaluation Metric For Machine Learning, Transactions of Machine Learning Research, 2023. [code]
Vendi Sampling For Molecular Simulations: Diversity As A Force For Faster Convergence And Better Exploration (Under review at Journal of Chemical Physics) [code – will be available after paper publication]
What sort of collaborations do you do, and what are you looking for?
I like to collaborate with people who have expertise that I don’t have because it is an opportunity to learn and have something unique accomplished that neither of the two collaborators could have achieved alone. So my favorite type of collaboration right now is collaboration with experimentalists, because then there’s an opportunity to actually get the structures we discover made and tested in the lab.
What's the hardest problem that you are facing at the moment?
There are many challenges that come with being a junior faculty working at the intersection of AI and science, coming from a machine learning background. One such challenge is finding a common language at the intersection of AI and science so as to disseminate our research to both ML and science communities.
What makes AI particularly well suited to designing nanoporous materials?
The process for discovering a new material is a very time-consuming process. The space of materials is a very large space and finding the material with the right property in this space may be a daunting task. I think AI is well-suited for materials design because it can [help] narrow down the search and find the right material. However, I think our current AI methodologies aren’t well-suited to problems in materials science off the shelf, which is why I am interested in developing novel AI techniques for the design of materials.
It is widely reported that you are the first Black faculty in Computer Science in Princeton’s history, the first Black woman tenure-track faculty in Princeton’s School of Engineering, and the second Black woman tenure-track faculty in Computer Science across the entire Ivy League. Where did you grow up, and how did you get exposed to science?
I grew up in Kaolack, a small town in Senegal. It is a ~3-hour drive from the capital Dakar. I was born and raised there and left to study abroad after I finished high school.
We learn science in high school in Senegal: chemistry, physics, biology, general science. So my first exposure was at school in Senegal. I was very fond of STEM subjects and philosophy.
Right before my senior year in high school, I was chosen to participate in a “girls’ excellence camp” organized by Dr. Cheick Modibo Diarra’s nonprofit organization at the time. It is only when I attended the camp and met Dr. Diarra and learned more about him that I really understood that one can make something out of being a scientist. It was all abstract to me before that. This is the reason why I set up the nonprofit The Africa I Know, so that kids back in Senegal and other countries in Africa can see themselves in the sciences and the possibilities that science and STEM in general offer, instead of being burdened by the stereotypes, often negative, associated with the continent.
Could you have made all of your contributions in your home country, or did you need to leave it? How did that make you feel then, and how does it make you feel now?
I don’t know what would have happened if I stayed in Senegal after I graduated from high school. That is a counterfactual I am comfortable with not knowing. I know that I left after graduating high school because I was yearning for more than universities in my country at the time could offer.
There were only two universities in Senegal when I finished high school: one in Dakar (Cheikh Anta Diop University) and one in Saint-Louis (Gaston Berger University). There was also at the time, and still to this day because our governments don’t invest much in education, a lot of instability in our education system due to strikes taking place during the year. I didn’t want to jeopardize my chance at a high-quality education and I was fortunate to have other options.
My dream is that Senegal and other African countries have a stable high-quality education system such as the ones in the West, so that an African student who decides to study abroad does so not because they have to but because they choose to.
Do you think the "brain drain" of AI professionals from Africa, South America, and other areas is a concern, or does it make sense for schools and companies in the global north to scoop up the very best, transport them to the United States and Europe, and give them huge resources to do the best that they can possibly do?
I think of the world as something that belongs to all of us that we all need to take care of, in collaboration. In that sense, I don’t believe in geography as a determining factor when it comes to participating in solving issues in different parts of the world. I believe in following the opportunity for acquiring knowledge wherever it is, as long as one is aware of the issues that are to be solved and takes part in tackling those issues however one can. This is why I don’t believe in brain drain. I think there’s no brain drain if there is awareness of the issues and a sense of responsibility toward the world as a whole. Anyone anywhere can bring positive change in any part of the world if only they choose to.