OpenAI just released Prism, a free tool that weaves ChatGPT directly into the text editor where scientists write papers. It's a deliberate move: if AI has reshaped how programmers work in 2025, OpenAI's betting that 2026 will be the year researchers stop treating ChatGPT as a curiosity and start treating it as part of the job.
The numbers suggest that shift is already underway. About 1.3 million scientists are already submitting over 8 million queries per week to ChatGPT on advanced scientific and mathematical topics. That's not dabbling. That's workflow integration happening in real time.
Prism runs on OpenAI's latest GPT-5.2 model, specifically trained to handle mathematical and scientific reasoning. Within the text editor, researchers can ask it to draft sections, summarize related papers, manage citations, convert handwritten equations into digital form, or work through a proof. The tool sits where the work already happens — no separate tab, no context-switching.
We're a new kind of news feed.
Regular news is designed to drain you. We're a non-profit built to restore you. Every story we publish is scored for impact, progress, and hope.
Start Your News DetoxKevin Weil, who heads OpenAI's science division, is clear about what this isn't: it's not a fully automated "AI scientist" that will suddenly announce a cure or a new theorem. That expectation misses the actual mechanism of scientific progress. What matters instead is what Weil calls "compounding acceleration" — thousands of researchers each saving a few hours per week on drafting, literature review, and notation. Multiply that across a global scientific community, and the cumulative effect reshapes what gets attempted and what gets finished.
That distinction matters because it's realistic. AI excels at the scaffolding of research — the repetitive, necessary work that creates the space for actual thinking. It's less about replacing the scientist and more about removing friction from the path they're already walking.
The timing is deliberate. 2025 proved that AI integration into existing workflows doesn't require perfection; it requires usefulness. Prism is betting that researchers will find the same value in having a capable assistant embedded in their writing process that developers found in code completion and debugging help.
