The size of chemical space is vast. This makes application of
first principles quantum mechanical and advanced statistical
mechanics sampling methods to identify binding motifs,
conformational equilibria, and reaction pathways extremely
challenging, even when considering better physical models,
algorithms, or future exascale computing paradigms. I will
describe the conceptual advantages of foundation models [1], and
the emergent properties that can arise as illustrated from large
language models for exploring chemical space of drug
molecules[2], chemical synthesis[3], transition metal
complexes[4], and the ability to predict Z-matrix geometries of
small molecules to proteins. I will also speak to the ethical
concerns of AI for the chemical sciences[5].
ย Teresa Head-Gordon is an international leader in theoretical
chemistry, creating methodological advances at the interface
between quantum and statistical mechanics, and AI and machine
learning. Her fundamental work includes theory and methods to
understand water and aqueous solvation, molecular interactions,
interfaces, chemical reactivity and (bio)catalysis, drug
discovery, and protein biophysics with notable impacts on
energy, environment, and human health applications. She is
Director of CalSolv at UC Berkeley, a Co-Director of the NSF
Molecular Sciences Software Institute, and has taken on many
leadership and advisory roles for various directorates within
NIH, DOE, NSF, and NAS panels and scientific decision bodies in
Europe. Honors include IBM SUR award; Schlumberger Fellow,
Cambridge University, UK; Fellow, AIMBE; Fellow, ACS; Fellow,
ReSolv German Center of Excellence, Humboldt Research Award, and
election to the International Academy of Quantum Molecular
Science. She has given multiple public and citizen lectures in
the US and Europe on the Ethics of Emerging Technologies: The
Era of Artificial Intelligence, a lecture taken from her popular
ethics course at UC Berkeley.