AI+Science Initiative at UChicago
Scientific discovery, and innovation have traditionally relied on four distinct elements: theory, observation, simulation, and practice to advance knowledge. However, a new engine of scientific knowledge generation is now emerging, fueled by extraordinarily powerful computers coupled with advanced instruments capable of collecting enormous volumes of high-resolution experimental data. Indeed, the challenge is now to inspect that data and to turn it into information, and subsequently into knowledge — a task humans can no longer perform unaided.
Off-the-shelf machine learning tools cannot fully extract the knowledge contained in these datasets, let alone generate new theories and propose future experiments. Only powerful machines and sophisticated algorithms informed by domain-specific constraints will advance the AI+Science revolution. Modern artificial intelligence and machine learning will fundamentally change scientific discovery. We are just beginning to understand the possibilities presented by an era where inference will no longer be limited by human experience. To advance this new era, we have the opportunity to make AI an integral component of the scientific method, guiding the construction of hypotheses, designing sequences of experiments, and analyzing data to develop new hypotheses, while advancing the foundations of AI.
The University of Chicago Data Science Institute’s AI+Science Initiative will lay the foundations for a new field of research, with cross-disciplinary teams of computer scientists, mathematicians, statisticians, engineers, physicists, chemists, biologists, geoscientists, and other domain scientists. AI-enabled scientific inquiry will allow us to discover new fundamental principles; accelerate the pace of scientific discovery in multiple fields, identify gaps in our knowledge, models, and understanding; and vastly expand the range of exploration and experimentation that can be subject to investigation. The impact of these developments on national competitiveness and technological preeminence cannot be overstated and, if properly applied, their relevance can be applied to some of the greatest challenges facing humanity.
The AI+Science initiative builds upon the work of the AI Joint Task Force, formed in 2019 by members of the University of Chicago, Argonne National Laboratory (ANL), Fermi National Accelerator Laboratory (FNAL), and the Toyota Technological Institute at Chicago (TTIC). The goal of this program is to encourage new interactions between the University, TTIC, and the national laboratories, provide unique educational experiences for students, and position the institutions to successfully secure future large-scale federally-sponsored research programs.
Scientists will use computational techniques to map the neural network of an octopus, the largest invertebrate brain, and better understand the fundamental principles of neuroscience.
Peter Littlewood (UChicago), Nicola Ferrier (Argonne), Bobby Kasthuri (UChicago)
Using machine learning and computer vision to classify cloud textures and patterns, and how those features have changed over time, in order to improve climate projections.
Rebecca Willett (UChicago), Ian Foster (Argonne, UChicago), Elisabeth Moyer (UChicago), Michael Maire (UChicago)
Deploying AI to automate the design, optimization and forecasting of cosmic survey experiments.
Brian Nord (Fermilab), Yuxin Chen (UChicago)
Using AI to find new, efficient ways of “recycling” CO2 into valuable products such as ethanol and ethylene, using clean electricity generated from the sun and wind.
Chibueze Amanchukwu (UChicago), Rajeev Assary (Argonne)
Designing new “system-on-a-chip” hardware to help researchers in data-intensive fields monitor data quality and detect promising results without interrupting the flow of data.
David Miller (UChicago), Nhan Tran (Fermilab), Andrew Chien (UChicago)
Exploring the mathematical limits of learned emulators for climate, astrophysics, and high-energy physics models and quantify key trade offs related to accuracy and speed.
Rebecca Willett (UChicago), Dana Mendelson (UChicago), Prasanna Balaprakash (Argonne), Jiali Wang (Argonne), Rao Kotamarthi (Argonne)
A machine learning pipeline that extracts the underlying principles of cell behavior from experimental data, offering new possibilities for the application of bioengineering in clinical diagnosis.
Margaret Gardel (UChicago), Vincenzo Vitelli (UChicago)
Using machine learning and statistical modeling to create thousands of synthetic microbiomes that could help safeguard human health and protect the environment.
Seppe Kuehn (UChicago), Arjun Raman (UChicago)
Developing new machine learning techniques to improve spectroscopy and discover new molecules present in your body, your microbiome, and the environment.
Eric Jonas (UChicago)
Using artificial intelligence (AI) to sift through thousands of molecules to identify target PFAS molecules and use these computational screens to guide the design and synthesis of new molecules that can remove them.
Junhong Chen, Andrew Ferguson, Eric Jonas, Stuart Rowan, and Rebecca Willett (UChicago) and Chris Benmore, Seth Darling, and Sang Soo Lee (Argonne)
Combining machine learning with high-throughput gene synthesis and rapid assay to learn nature’s rules for protein design and create synthetic proteins with elevated or new properties.
Andrew Ferguson (UChicago), Rama Ranganathan (UChicago)