An international team of experts in different fields and belonging to different research centers joined forces to develop a new artificial intelligence tool for scientific discovery using the same technology as ChatGPT.
While OpenAI’s chatbot works with words and phrases, these scientists’ AI — called Polymathic AI — is designed to process numerical data and physical simulations from all areas of science. The goal is to enable researchers to model all types of phenomena, from supergiant stars to Earth’s climate.
Polymathic AI principal investigator Shirley Ho, who works at the Center for Computational Astrophysics at the Flatiron Institute in New York, says the tool will change the way artificial intelligence and machine learning are used in the world of scientific research.
Numbers instead of characters
This is because, because Polymathic AI processes numbers as values instead of characters, it avoids the precision limitations associated with language models. Moreover, because this system will learn using data from a variety of sources in physics, astrophysics, and possibly chemistry and genomics, it will be able to apply this multidisciplinary knowledge to a wide range of scientific problems, overcoming the boundaries between disciplines that normally exist. in the world of research and in other artificial intelligence tools.
“Despite the rapid progress of machine learning in recent years in various scientific fields, in almost all cases machine learning solutions are developed for specific applications and trained on some very specific data; this creates boundaries both within and between disciplines, meaning that scientists who use AI for their research do not benefit from information that may exist in a different format or in a different field of research than their own, explains Francois Lanusse. cosmologist from the CNRS (France), in a statement with which he has the University of Cambridge. announced the Polymathic AI initiative.
Scientists using AI do not benefit from information that may exist in a different format or in a different field of research.
And the intent of the team of scientists building Polymathic AI is for their tool to enable leaps between disciplines. The project will “bring together many seemingly disparate subfields into something greater than the sum of its parts,” said Mariel Pettee, a postdoctoral researcher at Lawrence Berkeley National Laboratory.
“Our goal is to accelerate the development of versatile basic models designed for numerical data sets and scientific machine learning tasks. The challenge we accept is to create artificial intelligence models that use information from heterogeneous data sets and from different scientific fields that, as opposed to a domain like is the processing of natural language, they do not share a unified representation (i.e. text)”, the initiators of the initiative explained when announcing their project.
And they noted that the models they are developing can be used “as solid baselines, or researchers can refine them for specific applications; this approach has the potential to democratize AI in science by providing ready-to-use models that have stronger records (that is, prior knowledge) for general shared concepts such as causality, measurement, signal processing, and even more specialized shared concepts such as waves , something that would otherwise have to be learned from scratch.”
Read also
In this sense, one of the members of the group, Siavash Golkar, assures that the artificial intelligence they are training will be able to show common points or connections between different disciplines that may have been overlooked until now due to the increasing specialization within each. scientific fields, with research groups increasingly focusing on specific topics.
The Polymathic AI team includes experts in physics, astrophysics, mathematics, artificial intelligence, and neuroscience from the Simons Foundation and its Flatiron Institute, New York University, Cambridge University, Princeton University, and Lawrence Berkeley National Laboratory, among other institutions.
Read also
