OpenAI has introduced Prism, a free large language model powered tool designed to embed ChatGPT directly into the software that scientists use to write their papers. The company’s new in-house group, OpenAI for Science, built Prism to make Artificial Intelligence a central part of the scientific writing process, in much the same way that chatbots have become built into popular programming editors. Kevin Weil, who leads OpenAI for Science, drew a parallel with software development, saying, “I think 2026 will be for AI and science what 2025 was for AI in software engineering,” and describing a similar inflection point for research workflows.
OpenAI says that around 1.3 million scientists around the world submit more than 8 million queries a week to ChatGPT on advanced topics in science and math, and the company presents Prism as a response to that behavior and as a way to make Artificial Intelligence a core research tool. The product also serves as a strategy to keep scientists within OpenAI’s ecosystem at a time when there are many rival chatbots on the market. Some researchers already rely on OpenAI’s models: Roland Dunbrack, a biology professor at the Fox Chase Cancer Center, says, “I mostly use GPT-5 for writing code,” and notes that the models used to hallucinate references but do not seem to do that very much anymore. Nikita Zhivotovskiy, a statistician at the University of California, Berkeley, says GPT-5 has become important in his work, helping polish the text of papers, catch mathematical typos or bugs, provide useful feedback, and deliver quick summarization of research articles.
Prism follows products such as OpenAI’s Atlas, which embeds ChatGPT in a web browser, and similar office tools from Microsoft and Google DeepMind, but focuses on scientific writing in LaTeX. The tool incorporates GPT-5.2, described as the company’s best model yet for mathematical and scientific problem-solving, into an editor for writing LaTeX documents. A ChatGPT chat box appears at the bottom of the screen under the article view, allowing scientists to ask for help drafting text, summarizing related work, managing citations, converting photos of whiteboard notes into equations or diagrams, or exploring hypotheses and proofs. While Prism could save researchers significant time, it also raises concerns about adding to the volume of low-quality Artificial Intelligence generated content in science, especially amid hype around GPT-5’s problem-solving abilities and expectations for fully automated Artificial Intelligence scientists. Weil emphasizes that the goal is not a single dramatic breakthrough but broad, cumulative gains, saying, “I think more powerfully-and with 100% probability-there’s going to be 10,000 advances in science that maybe wouldn’t have happened or wouldn’t have happened as quickly, and AI will have been a contributor to that,” framing Prism as a tool for incremental, compounding acceleration rather than a standalone discovery engine.
