New Research Shows How AI Could Transform Math, Physics, Cancer Research, and More

New research shows how AI could transform math, physics, cancer research and more

By Deni Ellis Béchard edited by Clara Moskowitz

A new report from OpenAI and a group of external scientists shows how GPT-5, the company’s latest AI extended language model (LLM), can help with research from black holes to cancer cells to mathematical puzzles.

Each chapter of the article offers case studies: a mathematician or physicist stuck in a dilemma, a doctor trying to confirm a laboratory result. They all ask GPT-5 for help. Sometimes the LLM gets it wrong. Sometimes it finds a faster path to an already known outcome. But other times, with careful human guidance, it helps push the boundaries of what was previously known.

In an experiment investigating the behavior of waves around black holes, GPT-5 used mathematics to independently produce results that had previously been shown to be correct, demonstrating that it was capable of performing this level of scientific computing. In another project involving nuclear fusion, GPT-5 developed a model that accelerated research. “AI’s ability to dramatically reduce the time required for coding (compressing what would traditionally take the author days into minutes) has monumental implications for research practices,” says Floor Broekgaarden, an astronomer at the University of California, San Diego, who was not involved in the study.

On supporting science journalism

If you enjoy this article, please consider supporting our award-winning journalism by subscription. By purchasing a subscription, you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.

In another case, researchers studying immune cells used GPT-5 to interpret their data, and its explanation matched results already confirmed by the lab. “GPT-5 Pro can function as a true mechanistic co-investigator in biomedical research, compressing months of reasoning into minutes, uncovering non-obvious hypotheses, and directly shaping experimentally testable strategies,” Derya Unutmaz, the lead physician on the project, wrote in the article.

The paper also announces several new mathematical discoveries supported by GPT-5. Guided by human experts, it solved a long-standing problem posed in 1992 by mathematician Paul Erdös. It also produced a clearer rule showing the limits of how computer systems make decisions; discovered another rule for how certain small patterns appear in branching diagrams; and found a way to spot the secret structures of a network as it grows. The findings are modest but appear genuine, and each has been verified by human mathematicians.

“I had never seen anything so impressive [in math] “I suspect that LLMs will revolutionize the way theories are created, tested and improved.” He cautions, however, that AI tools still require a lot of incentive: “Humans are creative; The AI ​​is responsive. However, the pace of discoveries is expected to increase rapidly.”

Prithviraj Ammanabrolu, a computer scientist at the University of California, San Diego, who was not involved in the research, points out that the published work is more of a series of case studies than a scientific paper because it does not provide enough detail to repeat the experiments and does not offer a counterfactual analysis involving different approaches. Despite these limitations, AI’s ability to contribute to research “is still well ahead of what was possible just a year ago, so the rate of progress is quite high,” he says. “This shows future potential by allowing scientists to precisely blend relevant previous results and derive new insights in new ways.”

One of the strengths of GPT-5 is its ability to search large amounts of scientific literature. For a math problem listed online as unsolved, she identified a solution in an article from the 1980s. In another case, she found a few lines in a German newspaper from the 1960s that solved a problem. He easily overcame the language barrier and style differences between mid-century mathematical writing and contemporary approaches.

All of this might make it sound like GPT-5 is scientific genius, but the authors of the paper make it clear that this is not the case. In the right hands, it’s more of a quick, tireless assistant who has read an impossible number of documents and doesn’t bother reworking a calculation. But human judgment is not optional, they emphasize. Researchers also found that it was fake, and could misrepresent references, hallucinate non-existent articles, or fail to credit authors of real articles.

“Human expertise remains crucial,” says Broekgaarden. But AI “can take on myriad tasks—gathering data, summarizing research articles, and even performing complex calculations—that previously required significant time and effort on the part of researchers.”

Many ways in which AI will shape research remain to be discovered. New AI models are released every few months. If general-purpose chatbots that struggled with middle school math two years ago can now spot hidden structures in black hole waves and suggest new approaches to cell therapy, who knows what their successors will achieve?

It’s time to defend science

If you enjoyed this article, I would like to ask for your support. Scientific American has been defending science and industry for 180 years, and we are currently experiencing perhaps the most critical moment in these two centuries of history.

I was a Scientific American subscriber since the age of 12, and it helped shape the way I see the world. SciAm always educates and delights me, and inspires a sense of respect for our vast and beautiful universe. I hope this is the case for you too.

If you subscribe to Scientific Americanyou help ensure our coverage centers on meaningful research and discoveries; that we have the resources to account for decisions that threaten laboratories across the United States; and that we support budding and working scientists at a time when the value of science itself too often goes unrecognized.

In exchange, you receive essential information, captivating podcasts, brilliant infographics, newsletters not to be missed, unmissable videos, stimulating games and the best writings and reports from the scientific world. You can even offer a subscription to someone.

There has never been a more important time for us to stand up and show why science matters. I hope you will support us in this mission.