IIT Delhi's AI agent can conduct AFM experiments autonomously
A team from IIT Delhi, University of Jena, and Aalborg University has built an AI system called AILA that can run atomic force microscopy (AFM) experiments on its own.
AILA uses advanced language models to handle everything—calibrating the equipment, capturing images, analyzing results, and making decisions—without direct human intervention during experiments, though human oversight remains important for safety.
The research was published in Nature Communications.
How does AILA actually work?
AILA's setup includes a central planner that tells specialized agents what to do—one manages the microscope, another handles data crunching.
The team tested several popular AI models on 100 real lab tasks. AILA proved it could fine-tune settings, count graphene layers, measure friction, and analyze shapes, demonstrating the ability to perform these tasks autonomously, though with notable limitations in accuracy and reliability.
What do the researchers think?
Prof. Lothar Wondraczek says AI like AILA is meant to boost creativity for scientists—not replace them.
The project started when Prof. N M Anoop Krishnan spent time at Friedrich Schiller University Jena with support from the Alexander von Humboldt Foundation.
While multi-agent setups worked better than single ones, they were still pretty sensitive to changes in instructions—a reminder that even smart tech needs careful handling.