Autonomous or “self-driving” labs are an emerging technology where artificial intelligence guides the discovery process, helping design experiments or perfecting decision strategies.
While these labs have generated heated debate about whether humans or machines should lead scientific research, a new paper from Argonne National Laboratory and the University of Chicago Pritzker School of Molecular Engineering (UChicago PME) has proposed a novel answer: Both.
In a paper published in Nature Chemical Engineering, the team led by UChicago PME Asst. Prof. Jie Xu, who has a joint appointment at Argonne, outlined an “AI advisor” model that helps humans and machines share the driver’s seat in self-driving labs.
“The advisor will perform real-time data analysis and monitor the progress of the self-driving lab’s autonomous discovery journey,” Xu said. “If the advisor observes a decline in performance, it is going to prompt the human researchers to see if they want to switch the strategy, refine the design space or so on.”
Inspired by the software used to help investors trade stocks, the model leverages AI’s data-processing prowess but keeps decisions in the hands of experienced researchers accustomed to making real-time choices using limited datasets. By leveraging the strengths of each side, Xu said this collaborative model made the often-lengthy workflow of deigning and testing materials much more efficient.
“We promote human-machine collaboration to boost discovery together,” she said.
Co-corresponding author Henry Chan, a staff scientist at the Nanoscience and Technology division at Argonne, said the goal is not to put either AI or humans in charge, but to have each focus on what they do best.
“People have been focusing a lot on self-improving AI—AI that can modify its own algorithm, generate its own data set, retrain itself and all that,” Chan said. “But here we're taking a cooperative approach where humans can play a role in the process also. We want to facilitate the collaboration between human and AI to achieve co-discovery.”
Putting the AI advisor to work
The team applied the advisor model to work on an electronic materials challenge using the self-driving lab Polybot, located in Argonne’s Center for Nanoscale Materials, to study and design electronic material called a mixed ion-electron conducting polymer.
The polymer created through this merger of machine and human intelligence showed a 150% increase in mix conducting performance over those created through the previous cutting-edge technique—while also identifying two key factors for this improvement.
This advance in pure science will help future researchers better design such polymers, said UChicago PME Assoc. Prof. Sihong Wang, also a co-corresponding author on the new paper.
“For material science research, there are two related goals,” Wang said. “One is to improve the material’s performance or develop new performance. But to enable that, you need the second goal—a deep understanding about how different material design strategies, parameters and processing conditions will influence that performance. By making the entire space of the structure variation much larger, this AI model has helped to achieve two goals at the same time.”
The team next looks to improve the communication not from the AI, but to it, helping humans and software better advance science together.
“Currently, the interaction is mostly one-way. Information is coming from the AI advisor, then humans take optional actions,” Chan said. “In the future, we want a tighter integration between AI and humans, where the AI can learn from human actions and modify the way it thinks in subsequent iterations, modeling the way of human decision-making.”
Citation: “Adaptive AI decision interface for autonomous electronic material discovery,” Dai et al, Nature Chemical Engineering, December 18, 2025. DOI: 10.1038/s44286-025-00318-3
—Adapted from an article originally published on the UChicago PME website.
