Research | King-Smith Group

Synthesis in the Age of AI

We are at a critical junction in the field of synthetic chemistry. We can make beautiful and complex molecules. We can forge new carbon-carbon bonds through a plethora of methods. We can even do this through (somewhat) environmentally friendly means. But synthesis is still highly empircal, with routes forged through hundred or thousands of failed reactions.

Can we change this with machine learning and AI?

arrows symbolizing the areas of interest in reaction prediction: regioselectivity, yield, and enantioselectivity

3D blocks symbolizing the multi-dimensional vector space of molecular embeddings

A chemical flask to symbolize AI-guided synthesis

In the King-Smith Group, we investigate how machine learning can improve our own chemical intuition utilizing the rich landscape of deep learning and firmly grounded through in-house experimental validation. Our three areas of focus are currently reaction outcome prediction, molecular embedding design, and AI-guided synthesis.