AIchemy’s Monthly Webinar Series – May 2026

We are delighted to welcome you to our AIchemy Hub’s monthly webinar series.

This month’s talks:

Prof. Bao Nguyen – University of Leeds

Talk Title: Who’s learning from whom? Beyond the black boxes of chemical models.

Artificial intelligence and machine learning are now central tools for chemists seeking to predict molecular properties and reaction outcomes. Yet as these models grow increasingly sophisticated, their inner workings often remain opaque, and the chemical data they rely on—like all experimental data—can be noisy, sparse, or biased. In this talk, Bao will illustrate how we address these challenges in the context of solubility prediction: from handling imperfect datasets to building models that both perform robustly and provide trustworthy predictions on previously unseen data.

He will then show how the usual paradigm can be reversed. Rather than using algorithms solely to predict the results of complex reactions, we can use the data generated through Bayesian Optimisation to reveal mechanistic insights that would otherwise remain hidden. This shift—from prediction to understanding—opens new opportunities for rationally tackling selectivity problems in modern synthetic chemistry.

Nikola Radulov – University of Liverpool

Talk Title: FLIP: Flowability-Informed Powder Weighing

Autonomous manipulation of powders remains a significant challenge for robotic automation in scientific laboratories. The inherent variability and complex physical interactions of powders in flow, coupled with variability in laboratory conditions necessitates adaptive automation. We introduce FLIP, a flowability-informed powder weighing framework designed to enhance robotic policy learning for granular material handling. The core of the framework lies in using material flowability, quantified by the angle of repose, to optimise physics-based simulations through Bayesian inference. This yields material-specific simulation environments capable of generating accurate training data, which reflects diverse powder behaviours, for training “robot chemists”.  We demonstrate how FLIP integrates quantified flowability into a curriculum learning strategy, fostering efficient acquisition of robust robotic policies by gradually introducing more challenging, less flowable powders. We validate the efficacy of our method on a robotic powder weighing task under real-world laboratory conditions. Experimental results show that FLIP with a curriculum strategy achieves a low dispensing error of 2.12 +/- 1.53 mg, outperforming methods that do not leverage flowability data, such as domain randomisation (6.11 +/- 3.92 mg). These results demonstrate FLIP’s improved ability to generalise to previously unseen, more cohesive powders and to new target masses.