Science at the Frontier: How scientists use ChatGPT
OpenAI has published a piece titled "Science at the Frontier: How scientists use ChatGPT," which examines how more than 1,800 scientists apply ChatGPT in their research. The company frames the material around three themes: how the tool fits into everyday workflows, the productivity scientists report, and what they expect from it in the pursuit of discovery.
The figures and themes come from OpenAI's own account; the underlying survey detail, methodology and any quantitative results are not laid out beyond those headline points. Readers should treat the reported productivity effects as self-reported by the participating scientists rather than as measured outcomes from controlled studies.
This is best read as a signal of adoption rather than an evaluation of capability. Large language models can speed up literature triage, drafting and code, but studies of their use in science have documented risks around fabricated citations and overreliance, which makes verification practices central to any workflow.
Why it matters: research teams and the toolmakers serving them now have a sizeable user population to design around, which will likely push demand for provenance, reproducibility and audit trails in AI-assisted work. For developers building for scientific users, that suggests features for citing sources and checking outputs may matter as much as raw model quality.
Based on reporting from the original publisher. Visit the source for full context and later updates.
Publisher excerpt
Explore how more than 1,800 scientists use ChatGPT in research. See findings on daily workflows, reported productivity, and expectations for discovery.