All AI companies have been saying that they expect AI to greatly accelerate scientific research, and there are reports now coming in from the ground that this indeed seems to be happening.
British mathematician Timothy Gowers, who won the Fields Medal in 1998 for research connecting the fields of functional analysis and combinatorics, has said that GPT-5 took 20 seconds to come up with a proof that would’ve taken him more than an hour. He says we’re now in an era where AI can greatly speed up research, but still needs humans for supervision.
“I crossed an interesting threshold yesterday, which I think many other mathematicians have been crossing recently as well,” Gowers said on X. “In the middle of trying to prove a result, I identified a statement that looked true and that would, if true, be useful to me,” he added.
“Instead of trying to prove it, I asked GPT5 about it, and in about 20 seconds received a proof. The proof relied on a lemma that I had not heard of (the statement was a bit outside my main areas), so although I am confident I’d have got there in the end, he time it would have taken me would probably have been of order of magnitude an hour (an estimate that comes with quite wide error bars). So it looks as though we have entered the brief but enjoyable era where our research is greatly sped up by AI but AI still needs us,” he said.
“I’ve used LLMs many times, but this felt like the moment it becomes a regular part of my workflow at the thinking stage,” he said.
There are already indications that AI systems are able to perform scientific tasks. OpenAI and Google have won Gold medals at the International Math Olympiad, and they’ve also won gold medals at top coding competitions. Serious math researchers like Timothy Gowers are also saying that AI is helping them speed up their research.
And it’s this ‘speed up’ that seems to be at the heart of the predictions about how AI will unleash a new era of science and discovery. AI researchers can be created in large numbers, and they’ll never tire or want vacations. These AI researchers can be tasked with solving difficult scientific problems, and as they make new discoveries, the pace of change will only accelerate, leading to rapid technological advancements. Anthropic CEO Dario Amodei has previously said that AI will lead to a century’s worth of scientific progress in 5-10 years, Google DeepMind’s Demis Hassabis has said that AI could potentially cure all diseases. And with reports coming in that professional mathematicians are already saving hours with the use of AI, this new era of scientific discovery might well be closer than most people expect.