IT Brief UK - Technology news for CIOs & IT decision-makers
United Kingdom
OpenAI study finds scientists use ChatGPT to speed research

OpenAI study finds scientists use ChatGPT to speed research

Tue, 17th Nov 2026 (Today)
Mara Sugue
MARA SUGUE News Editor

OpenAI has published research on how scientists use ChatGPT in their work. The study examined messages from 1,819 researchers and survey responses from 251 participants.

OpenAI's Economic Research team analysed more than 580,000 scientific messages from users in its ChatGPT for Academic Researchers summer pilot. It found that scientists used the tool across a wide range of tasks, including reviewing literature, analysing evidence, writing software and developing research ideas.

Participants estimated that doing the same work to the same standard would have taken 1.9 times as long without AI. Nearly all respondents who answered a question on discovery timelines said they expected AI to shorten the time needed for breakthroughs in their field.

Daily use

AI has become part of scientists' day-to-day work, according to the study. The largest shares of scientific messages related to reading and learning at 28%, writing and communicating at 27%, and analysing evidence and theories at 17%.

Developing ideas accounted for 14% of scientific messages in the dataset, while writing research software made up 5%. Smaller shares related to teaching and mentoring, research administration, reviewing research and gathering data.

The figures also showed how researchers approached idea generation. In 35% of ideation-related exchanges, scientists brought a specific question for ChatGPT to work with. In 36%, they arrived with an idea and asked the system to refine it. In 23% of those messages, scientists asked ChatGPT to suggest a research question.

Follow-up checks

The study also found that researchers often stayed involved after delegating tasks to the chatbot. OpenAI measured how often scientists followed up to clarify, check or revise responses across different research activities.

Writing and communicating produced the highest rate of follow-up requests at 38%. Writing research software and analysing evidence and theories each stood at 34%, while reviewing research was 32% and developing ideas was 31%.

Those figures suggest scientists were not simply handing over work and accepting the first answer. Instead, the study points to a pattern of researchers using the system iteratively across several stages of the research process.

Discovery timelines

Among the 250 respondents who answered a question on discovery timelines, 98% said AI would speed up discovery in their field. Participants were asked to imagine a discovery that would take 10 years without AI and estimate how long it would take with AI.

Across all fields, 22% said such a discovery could happen within three years with AI, and 50% placed the timeline at four to seven years. A further 19% chose eight to nine years, while 7% said it would take just under 10 years and 2% expected no acceleration.

Responses varied by discipline. In engineering, 59% of respondents selected four to seven years. In maths and statistics, 31% chose within three years. Life and health sciences had a comparatively higher share of respondents selecting eight to nine years.

OpenAI also asked scientists who expected faster discovery to explain why. Of the 367 reasons identified in 224 written responses, 52% said AI would help get research done faster, 23% said it could help find more promising research questions, and 14% said it could support work on a wider range of questions.

Privacy method

To analyse usage while limiting exposure to sensitive research material, OpenAI used automated classifiers to sort messages by activity type rather than having the research team read the underlying conversations or the chatbot's replies. Data from the ChatGPT for Academic Researchers programme is not used by default to train its models.

The research covered 580,186 scientific messages from 1,819 scientists using the academic plan, while the survey results were based on 251 responses. OpenAI noted that the number of answers varied by question, and some respondents selected multiple scientific fields, meaning the field groups overlapped.

OpenAI also said it is expanding ChatGPT for Academic Researchers with free access for more than 200,000 researchers worldwide. "Tools like ChatGPT are helping us accelerate scientific discovery. It's really a playground for a scientist like myself," said César de la Fuente, Antimicrobial Researcher.