An M.I.T. Ph.D. student and two University of California system cryptographers used GPT-5.6 Sol Ultra in different ways, raising new questions about independent discovery and scientific credit
Seems the proof more or less already existed and the LLM helped the researchers find it and demonstrate it applied to the problem they were researching? Basically doing grunt work literature searches that grad students might have done in the past.
Good point. The article does talk about how
Ragavan failed to find the solution a couple of years earlier, but the AI was able to find Broadbent’s paper that "gave it on a silver platter.”
Almost seems like the “smarter” people like Broadbent come up with the more novel revolutionary ideas, then others find ways of applying it to the rest of the world. And AI is able to do the latter as well.
Seems the proof more or less already existed and the LLM helped the researchers find it and demonstrate it applied to the problem they were researching? Basically doing grunt work literature searches that grad students might have done in the past.
Good point. The article does talk about how Ragavan failed to find the solution a couple of years earlier, but the AI was able to find Broadbent’s paper that "gave it on a silver platter.”
Almost seems like the “smarter” people like Broadbent come up with the more novel revolutionary ideas, then others find ways of applying it to the rest of the world. And AI is able to do the latter as well.