Company Updates

OpenAI study: scientists say AI saves 1.9x time on research tasks

Share
OpenAI study: scientists say AI saves 1.9x time on research tasks

OpenAI's analysis of 1,819 researchers shows AI tools cut the time needed for core research tasks by nearly half, with coding efficiency gains outpacing other activities. The company plans to scale its academic program to reach up to 200,000 scientists.

TL;DR

  • OpenAI's study of 1,819 scientists found AI tools reduced research time by 1.9x on average, with coding tasks 2.83x faster.
  • The company is expanding its academic pilot to cover up to 200,000 researchers, up from 150,000 in the current round.

What happened

OpenAI published findings from a study tracking 1,819 scientists using its AI tools between August 20 and September 22, 2026. The analysis covered 580,186 scientific messages, with reading and writing accounting for over half of all interactions.

Surveyed researchers estimated that without AI, equivalent work would take 1.9 times as long, with coding tasks showing the largest time savings at 2.83x. Research ideation followed at roughly twice the time required.

The study also found that fully autonomous research accounted for just 0.4% of messages, though 15% of participants tried at least one autonomous request. Older researchers spent more time on teaching and mentorship, while younger researchers focused more on coding.

OpenAI is expanding its academic pilot program, opening 50,000 grants on November 23 to cover up to 150,000 researchers. The program now includes ChatGPT Plus access, replacing the previous Pro tier used in the pilot.

Why it matters

For AI startups and developers, the study highlights where AI tools deliver the most immediate value in research workflows—particularly coding and drafting—while also underscoring the limits of full automation. The 0.4% share for autonomous research suggests most scientists still rely on AI as a productivity aid rather than a replacement.

Investors and academic institutions should note that time savings do not automatically translate to higher-quality research. OpenAI's findings indicate that about a third of requests required follow-up checks or revisions, meaning AI outputs still demand human oversight.

The expansion of OpenAI's academic program to 200,000 researchers could accelerate adoption of AI tools in science, but the study's limitations—such as self-reported data and a focus on OpenAI's own tools—mean broader trends remain uncertain.

For founders building AI tools for researchers, the data points to clear opportunities in workflow integration, particularly for coding and writing assistance, but also highlights the need for features that support verification and iterative refinement.

Key facts

  • OpenAI's study tracked 1,819 scientists using its AI tools, analyzing 580,186 scientific messages from August 20 to September 22, 2026.
  • Surveyed researchers estimated that without AI, equivalent work would take 1.9 times as long, with coding tasks 2.83x slower.
  • Reading and writing accounted for 27.7% and 26.9% of scientific messages, respectively, while coding made up 13.8%.
  • Fully autonomous research accounted for 0.4% of messages, though 15% of participants tried at least one autonomous request.
  • OpenAI's academic pilot program is expanding to cover up to 200,000 researchers, up from 150,000 in the current round.
  • The study found that 51.8% of respondents cited limited time or staffing as a constraint, while 43.8% pointed to funding.
  • Older researchers (40+) devoted more time to teaching and mentorship, while younger researchers (18-29) focused more on coding (17.9% vs. 9.8%).
  • About a third of AI requests required clarification, checking, or revision within a week, according to OpenAI's findings.

Context

AI tools are increasingly integrated into scientific workflows, with companies like OpenAI and Anthropic positioning their models as assistants for research tasks rather than replacements for human expertise.

The study's focus on time savings aligns with broader industry trends showing AI's impact on productivity in knowledge work, though questions remain about how these gains translate to scientific rigor and reproducibility.

OpenAI's academic program expansion reflects a growing trend of AI companies investing in research communities to drive adoption and gather real-world usage data.

Topics

Related coverage

Join the discussion

Have a take on this story? Weigh in with our community on Facebook.

💬 Discuss on Facebook →