VMTech
Discuss a project →

OpenAI survey shows scientists adopting ChatGPT across research

OpenAI survey shows scientists adopting ChatGPT across research

OpenAI examines ChatGPT use in scientific research

OpenAI’s Economic Research team has published findings from the ChatGPT for Academic Researchers summer pilot, examining more than 580,000 scientific messages from 1,819 scientists and surveying 251 participants. The study describes how researchers used ChatGPT across work including literature review, evidence analysis, coding, idea development and communication.

Among the headline results, 68% of scientists used ChatGPT to develop research ideas. Surveyed participants estimated that completing comparable work at the same quality would have taken 1.9 times as long without AI. Of the 250 respondents who answered a question about discovery timelines, 98% expected AI to accelerate discovery in their field.

Ideation is collaborative rather than fully delegated

OpenAI separated messages about research ideas into several patterns. Scientists arrived with a specific question 35% of the time, while 36% of messages involved researchers bringing an idea and asking ChatGPT to refine it. In 23% of cases, scientists asked the system to suggest a research question; 6% could not be assigned to the other categories.

The analysis also found that scientists remained engaged after asking ChatGPT for help. Researchers followed up to clarify, check or revise responses across different activities. That pattern frames the product as part of an iterative workflow, not as an independent actor making scientific decisions.

This use case connects with ChatGPT Work and supervised multi-step workflows as organizations consider systems that can carry work through multiple steps while retaining human review and accountability.

Speed is only one expected benefit

OpenAI asked respondents who anticipated faster discovery to explain why. Of 367 reasons identified in 224 written responses, 52% concerned completing research faster. Another 23% cited finding more promising research questions, and 14% referred to pursuing a broader range of questions. The remaining 11% fell into other categories.

The company says the findings do not establish that these expectations will translate into faster discovery. It notes that progress will still depend on the funding, staff and experimental capacity required to test ideas. To protect privacy in the study, OpenAI used automated classifiers to categorize research activity and did not read the underlying participant messages or ChatGPT responses.

What research organizations can take from the study

Alongside the report, OpenAI said it is expanding ChatGPT for Academic Researchers with free access for more than 200,000 researchers worldwide. For research organizations, the practical implication is to evaluate AI where it can support exploration, analysis and drafting, while preserving scientist-led verification and ensuring that the resources needed to test results remain in place.

#chatgpt#scientificresearch#artificialintelligence#researchtools
Open analytics
On the site 3 views
min read 3 16.11.2026
Instagram

OpenAI survey shows scientists adopting ChatGPT across research

Open the post on Instagram ↗