OpenAI Claims Scientists Are Increasingly Using ChatGPT as a Research Collaborator

OpenAI has described a growing role for artificial intelligence (AI) systems as collaborators in scientific research, saying tools such as ChatGPT can help researchers to advance on complex problems across disciplines including mathematics, physics, and chemistry and biology. The 20-page report cites that scientists are increasingly using AI models to support literature synthesis, data interpretation and experiment planning, and emphasizes the fact that advancement in AI’s research capabilities could accelerate discovery and productivity in science and engineering.

OpenAI Pitches ChatGPT as a Scientific Collaborator

A report – which was exclusively published with Axios, quotes anonymised usage data that millions of scientists, engineers and mathematicians are already asking more sophisticated questions as well as using AI to support scholarly work (ranging from writing scientific text to debugging code and planning experiments). Researchers primarily use AI for writing and communication tasks, according to OpenAI’s analysis, while less people use it for rigorous analysis and calculations, suggesting that there is an opportunity for deep integration of AI in the research process.

OpenAI said that “Anonymised ChatGPT conversations between January and December 2025, averaging week-wide message count on advanced science and mathematics related topics increased by about 47 percent year-on-year (YoY)” in an effort known as average weekly messages. They said the messages were up from 5 to 6 in absolute numbers. To nearly 8, 7 million or 7million to . 4 million,. The AI giant said in January 2026 that “close to 1” was a synonym of the word “synonymous phraser”. In science and mathematics, 3 million weekly users were talking about advanced topics in science & math.

The use of “AI” as a scientific collaborator is growing, and we’re seeing its influence in real-world research contexts. A more advanced reasoning system is being used by researchers to advance in the field of open problems, interpret complex data and iterate faster in experimental work. That usage has been growing rapidly over the last year, with results starting to appear across fields. Kevin Weil, VP of OpenAI for Science, said ‘We’re still early but the pace of adoption and quality of the work mean science is entering a new acceleration phase.

This idea of a collaborator for AI matches with projects like OpenAI for Science, based on the programme that seeks to connect scientists and mathematicians with tools designed to link researchers and mathematics in order to accelerate research workflows from literature analysis to modelling and simulation. Its goal is to create systems that integrate naturally into scientific practice, which help researchers explore ideas and test hypotheses faster and unlock discoveries that may take years under traditional methods,” OpenAI said.