AI Helps Scientists Find Plants That Pull Nickel From Soil

A crop better known as a cover plant is being tested as a possible tool for recovering nickel from soil—with an artificial intelligence system helping researchers decide which varieties perform best. According to Phys.org, scientists at the U.S. Department of Energy’s Oak Ridge National Laboratory studied 12 lines of pennycress collected from different growing regions. The plants were grown in soils containing varying concentrations of nickel, a metal used in lithium-ion batteries, stainless steel and superalloys. The experiment involved 360 plants observed inside Oak Ridge’s Advanced Plant Phenotyping Laboratory. Automated conveyors moved the plants past high-resolution cameras three times a day. The system recorded characteristics including leaf color, plant structure, growth, stress responses and mineral content, producing more than 24,000 observations. Researchers then used the Orchestrated Platform for Autonomous Laboratories, or OPAL, an agentic AI system developed through a collaboration involving several national laboratories. Unlike a chatbot designed to answer one question, the system can plan tasks, run code, retrieve results, identify patterns and suggest follow-up experiments while keeping human researchers involved. The AI identified pennycress varieties that tolerated nickel particularly well, assessed growth and stress resilience, and highlighted early traits that could predict how much nickel a plant would ultimately accumulate. It also flagged unusual results for researchers to review. The time savings were substantial. Phys.org reported that analyzing more than 1,000 plant traits previously required hundreds of hours of manual effort but could be completed through interaction with the AI in a few minutes. In a separate comparison, two researchers spent six hours recording 10 traits at one timepoint by hand, while the automated system could reproduce that work in less than a minute and analyze many more traits across the experiment. The broader approach is known as phytomining: growing and harvesting plants that absorb metals from soil, then processing the plant material to recover them. The method could be used on low-grade mineral deposits, marginal agricultural land and areas undergoing rehabilitation, according to the report. It is not yet presented as a replacement for conventional mining. Oak Ridge’s next steps include combining plant research with AI-designed proteins and microbes intended to help release critical minerals from soil and improve their uptake. The researchers emphasized that the AI assists with analysis and experimental planning, while scientific interpretation and decisions remain under human control.