Today, 25 September, ARIA held its AI Scientist Knowledge Exchange Meeting in a hybrid format at its headquarters in The Jellicoe, King’s Cross, London, with participation both in person and online. The meeting brought together teams selected through ARIA’s AI Scientist call to present project progress, discuss current scientific and technical bottlenecks, and compare approaches to AI-enabled scientific discovery.
Prof. Dr. Andrey Ustyuzhanin, Principal Investigator at Constructor Labs, represented MIND-MATTER: AI-Driven Discovery of Self-Learning Materials and presented the project’s current progress and findings.
Led by Constructor Labs in collaboration with the University of Twente and Constructor Tech, MIND-MATTER applies an AI Scientist workflow to the study of reconfigurable nonlinear processing units. The project uses AI-guided hypothesis generation, experimental design and data analysis to investigate how physical charge-transport mechanisms determine computational behaviour in self-learning neuromorphic materials.
The discussion focused on results, methodological challenges and the practical limits of current AI Scientist systems across the participating projects.
