- Scientists have developed a new AI tool to accelerate scientific discoveries
- LLM4SD explains the reasoning behind its predictions, for transparency
- Instead of replacing standard automatic learning models, LLM4SD improves them
An Australian research team led by Monash University has created a generative AI tool designed to accelerate scientific discoveries. Called LLM4SD (Large Model 4 Scientific Discovery), the open source tool recovers information, analyzes the data, then generates hypotheses from it.
Although the LLMs are used in natural sciences, their role in scientific discovery remains largely unexplored, and unlike many validation tools, LLM4SD explains its reasoning, making its predictions more transparent (and, hope, reducing hallucinations).
Doctoral student Yizhen Zheng from the Data Sciences Department of Monash University explains: “Like Chatgpt writes tests or solves mathematical problems, our LLM4SD tool reads decades of scientific literature and analyzes laboratory data to predict how molecules behave-answer questions like:” Can this medicine cross the protective barrier of the brain? ” Or “will this compound dissolve in the water?” “”
Simulation of scientists
The LLM4SD has been tested on 58 research tasks through physiology, physical chemistry, biophysics and quantum mechanics, and has surpassed the main scientific models, improving precision up to 48% in the forecasting of quantum properties crucial for the design of materials. Zheng said: “In addition to outperforming current validation tools that work as a” black box “, this system can explain its analysis process, its predictions and its results using simple rules, which can help scientists trust and act on his ideas.”
Doctoral student Jiaxin Ju of Griffith University said: “Rather than replacing traditional automatic learning models, LLM4SD improves them by synthesizing knowledge and generating interpretable explanations”.
The team considers the tool to “essentially simulate scientists”. Professor Geoff Webb of Monash University stressed the importance of the role of AI in research. “We are already entirely immersed in the era of the generator and we must start to exploit this as much as possible to advance science, while ensuring that we develop it ethical,” he said.
Research, published in Nature machine intelligence and available to display on the Arxiv pre-print server,, was a collaboration between the Faculty of Information Technology at Monash University, the Monash Institute of Pharmaceutical Sciences and the Griffith University.
