AI for chemistry: a new research frontier at Amsterdam Science Park

The urgency is clear. Developing materials that capture CO₂ from the atmosphere, storing energy from wind and solar power, designing next-generation pharmaceuticals, and building sustainable alternatives to fossil-based chemicals, all of these depend on chemistry moving faster and smarter than it currently can. Ensing argues that AI is the key enabler. By automating laboratory experiments, accelerating complex quantum-chemical simulations, and generating entirely new molecular structures, AI-driven research can compress timelines that once took years into a fraction of the time.

From simulation to discovery

At HIMS, Ensing and his colleagues are developing AI models that go well beyond pattern recognition. One line of work uses Bayesian optimisation to automate and guide experiments,  a method that identifies optimal reaction conditions efficiently, even when only limited data is available. Autonomous systems built on this approach are already operational in Amsterdam laboratories, including work by dr. Bob Pirok on chemical analysis and prof. dr. Timothy Noël on organic synthesis.

A second focus area applies machine learning to computational chemistry. By training neural networks on quantum-chemical calculations, Ensing’s group can predict molecular properties at a speed up to a thousand times faster than traditional methods, while maintaining comparable accuracy.

The third and most transformative development is generative AI for chemistry: models that can design entirely new molecules from scratch, guided by desired properties. Just as large language models generate text, these systems learn the structural grammar of molecules and use it to propose new compounds: candidates for better medicines, improved battery materials, or more effective catalysts.

Read the original article on the HIMS website