Discover how autonomous AI agents operate lab equipment analyse results and reduce human supervision in complex scientific experiments.
Modern scientific instruments generate large volumes of data and require constant human supervision to tune parameters, run experiments and interpret results. This dependence on skilled operators slows research progress and limits access to advanced experimentation, particularly in complex nanoscale studies.
Addressing this challenge, researchers from IIT Delhi, in collaboration with teams from Denmark and Germany, have developed an AI agent called AILA (Artificially Intelligent Lab Assistant) that can autonomously conduct real scientific experiments.
AILA is designed to operate physical laboratory equipment rather than function only as a digital assistant. The system was tested on an Atomic Force Microscope (AFM), an instrument used to analyze materials at nanoscale resolution. The AI agent controlled the microscope, adjusted experimental parameters in real time, executed measurement tasks and analyzed results without human intervention.
Unlike traditional automation scripts, AILA uses large language model based decision making to adapt during experiments. The study found that while many AI models perform well on theoretical science tasks, only a subset could reliably manage real world experimental workflows that require continuous feedback and adjustment. Although some AI agents deviated from instructions during experiment, highlighting the need for strong safety controls before wider deployment in laboratories.
Prof. N. M. Anoop Krishnan, IIT Delhi, says, “Earlier AI could help write about science. Now it can design experiments, run them, collect data and interpret results.” The development aligns with India’s AI for Science initiatives and ongoing investments in AI driven research infrastructure.
