Artificial intelligence is increasingly being used as a research partner for mathematicians and scientists, according to a new OpenAI report provided to Axios, UNN writes.
Details
OpenAI claims that AI can make scientists more productive by increasing the volume of research that can be conducted, ultimately leading to more life-saving breakthroughs.
According to the OpenAI report, an internal analysis of a random sample of anonymized ChatGPT conversations from January to December last year showed:
- the average weekly number of messages on “advanced STEM topics” increased by almost 47% over the year;
- as of January this year, almost 1.3 million users discuss “advanced STEM topics” weekly, with an average of 8.4 million ChatGPT messages on these topics.
Topics, as indicated, include mathematics, physics, chemistry, biology, and engineering for graduate students and researchers.
Among the users and messages considered by OpenAI, ChatGPT was most often used for advanced research in computer science, data science, and artificial intelligence.
"More and more researchers are using advanced reasoning systems to solve open problems, interpret complex data, and accelerate iterations in experimental work," said Kevin Weil, OpenAI's Vice President of Science, in the report.
"We are still in the early stages, but the pace of adoption and the quality of work indicate that science is entering a new phase of acceleration," he noted.
According to the report, most scientists and engineers use ChatGPT for writing and communication. The smallest proportion uses it for analysis and calculations.
GPT-5.2 has now "moved from a competitive level towards mathematical research," according to the report, with most users turning to it for structural equation models.
The report also shows frequent use of ChatGPT for computational chemistry and elementary particle physics, among other types of work in biology, chemistry, and physics.
"OpenAI encourages policymakers to expand the use of artificial intelligence in science and research, including scaling up AI skills training, opening access to data and advanced AI technologies to more people, and modernizing AI infrastructure," the publication writes.