The system effectively illustrated Kepler’s third law of planetary motion, Einstein’s relativistic time-dilation principle, and Langmuir’s equation for gas adsorption.
AI-Descartes, the newest AI scientist, has been able to replicate the Nobel Prize-winning research by employing symbolic regression and logical reasoning for deriving accurate equations. This system works efficiently with real-world data and small datasets; its plans include automating the production of underlying theories.
In 1918, an American chemist, Irving Langmuir, wrote a paper discussing the actions of gas molecules adhering to a solid surface. His theory that solids have distinct areas for gas molecules to occupy, combined with the findings of his thorough experiments, enabled him to craft equations illustrating how much gas will stick according to pressure.
About a century after its initial discovery, AI scientists from IBM Research, Samsung AI, and the University of Maryland Baltimore County have replicated an important part of Langmuir’s Nobel Prize-winning research.
AI working as a scientist could reproduce Kepler’s third law of planetary motion. This law can calculate the time it takes one celestial body to make an orbit around another based on the distance between them. AI also made an accurate approximation of Einstein’s relativistic time-dilation law, which states that time slows down for objects in rapid motion.
Reasoning Tool Powered By Machine Learning
“AI-Descartes” is the latest AI scientist to be developed by researchers, alongside AI Feynman and other similar computing tools. These systems are based on symbolic regression, which works by finding equations that correspond with given data. This technology has been created to accelerate scientific progress.
When provided with basic mathematical operations, like addition, multiplication, and division, the systems can discover hundreds of potential equations that best represent the relationships within the data. This number can even climb up to millions.
Cristina Cornelio, a research scientist at Samsung AI in Cambridge, England, who is the first author of the paper, states that AI-Descartes has several benefits compared to other systems. However, its most remarkable feature is its capability to think logically. It can determine which equations best fit the existing scientific theory when multiple potential equations match the data well.
The system is set apart from other “generative AI” programs, such as ChatGPT, due to its capacity for rational thought. These other programs have significant language models, but their analytical capabilities are limited, and they often make mistakes with simple math calculations.
Cornelio says:
“In our work, we are merging a first-principles approach, which has been used by scientists for centuries to derive new formulas from existing background theories, with a data-driven approach that is more common in the machine learning era,”
“This combination allows us to take advantage of both approaches and create more accurate and meaningful models for a wide range of applications.”
A reference to the 17th-century mathematician and philosopher René Descartes is made in the name AI-Descartes. Descartes believed that scientific discovery was enabled by a few fundamental physical laws which could be deduced through logical reasoning.
Real-World Data Solutions With SUITED Platform
This system is especially effective on actual, noisy data, which can challenge traditional symbolic regression programs. These programs may miss the true signal due to their attempts to fit every minor data detail. Additionally, it performs admirably even when given a small amount of data – producing dependable equations with as little as ten points.
A barrier to the widespread use of AI-Descartes for frontier science might be the requirement to assign existing theories to open scientific questions. Nevertheless, efforts are underway to develop datasets containing measurement data and a related background concept for their system to be further improved and tested on new grounds.
They would also like to eventually teach computers how to peruse scientific documents and autonomously devise the fundamental principles behind them.
Co-author Tyler Josephson, an assistant professor of Chemical, Biochemical, and Environmental Engineering at UMBC, states this.
Tyler Josephson says:
“In this work, we needed human experts to write down, in formal, computer-readable terms, what the axioms of the background theory are, and if the human missed any or got any of those wrong, the system won’t work,”
“In the future,”
“We’d like to automate this part of the work as well, so we can explore many more areas of science and engineering.”
Josephson is driven by utilizing AI tools to progress chemical engineering, which is the focus of his research.
The team hopes their AI-Descartes will emulate the real person and motivate a new scientific research method.
Cornelio says:
“One of the most exciting aspects of our work is the potential to make significant advances in scientific research,”
AI-Descartes represents a scientific renaissance in artificial intelligence, integrating philosophical principles with AI development to create more advanced and human-like AI systems. The potential applications of AI-Descartes are vast, and its interdisciplinary approach has the potential to shape the future of AI research and development. As we continue to advance the field of AI, AI-Descartes serves as a promising direction for unlocking new possibilities and realizing the vision of artificial general intelligence.
Source: SciTechDaily
