Technology / Science

Add as a preferred source on Google

OpenAI Survey Finds 97.6% of Respondents Expect AI to Speed Discovery

The figure came from a hypothetical 10-year timeline question; the study measured expectations and self-reported estimates, not observed acceleration.

Updated 2 min read
A researcher consults a laptop beside a microscope in a San Francisco laboratory. / EconoTimes.ai
A researcher consults a laptop beside a microscope in a San Francisco laboratory. / EconoTimes.ai

OpenAI said 97.6% of 250 survey respondents expected artificial intelligence to bring an important discovery in their field sooner, a forecast that does not show whether AI has accelerated research.

Respondents imagined that the discovery would take 10 years without AI. Of them, 21.6% expected AI to move the timeline forward by seven to 10 years, while 72% anticipated a gain of at least three years. OpenAI rounded the 97.6% result to 98% in its public summary.

The survey had 251 participants, with 250 answering the discovery-timeline question. OpenAI's report, Science at the Frontier: How Scientists Use ChatGPT, is dated Nov. 6; the company published its summary Nov. 16. The report also analyzes 580,186 scientific messages from 1,819 participants in ChatGPT for Academic Researchers, collected from Aug. 20 through Sept. 22.

Survey participants estimated that, without AI, completing the same amount of work at the same quality would take 1.9 times as long. That self-reported estimate is averaged across eight scientific workflows.

In the message analysis, reading and knowledge acquisition accounted for 27.7% of the average scientist's scientific messages, while writing and communication made up 26.9%. Requests for end-to-end autonomous research represented 0.4%. A separate follow-up measure averaged 31.8% at the scientist level: it counted when the next scientific-task message in the same conversation, within seven days, sought clarification, checking or revision.

Researchers also reported limits on their work. Limited time or staffing was selected by 51.8% of respondents, limited funding by 43.8% and difficulty validating or reproducing results by 25.5%. Participants could choose up to three constraints. The report points to broader investment in science as a possible need.

The findings may not reflect scientists outside the program: participants had applied for expanded access and opted into a survey about AI. The report characterizes AI use through message records, productivity through participants' estimates and discovery through expectations, so it does not establish that AI caused faster discoveries.

OpenAI said 20,000 researchers in 33 countries had joined the program since its July pilot. Applications for the next round are scheduled to open Monday, Nov. 23, with 50,000 new grants to be awarded on a first-come, first-served basis. The company says it aims to reach up to 200,000 researchers.

Related articles

View all