The Royal Swedish Academy of Sciences awarded this week not one, but two of the most prestigious science awards in the world to artificial intelligence researchers. The Nobel Prize in Physics went to physicist John Hopfield and computer scientist Geoffrey Hinton, for their contributions to the neural networks that power today’s most powerful AI models. In Chemistry the prize went to David Baker, Demis Hassabis and John Jumper, the latter two having developed the AlphaFold AI model that definitively mapped the proteins known to humankind. Hinton himself not-so-subtly hinted in an interview with The New York Times that the committee was trying to send a message about the importance of the technology. But you don’t need to speculate about the committees’ intentions to realize that the two prizes represent, amid a sea of corporate hype and flashy product rollouts, a rare, concrete truth about AI — that it’s already utterly transformed the sciences, with profound implications for the research community (and, ahem, those who fund it, both in government and the private sector). “The obvious explanation is that this is a sign the committees were feeling behind the times, in a world where AI dominates the science headlines but the Nobels don't have an obvious way to recognize that,” Helen Toner, a former OpenAI board member and current director of strategy at Georgetown University’s Center for Security and Emerging Technology, told DFD. It’s common for the Nobels to identify a big leap in science, and then find ways to allocate credit for it. It is slightly less common, however, for the Nobels to award a tool — and the breadth of what they did this week tells us something bigger about AI’s disruptive capacity. Rohit Krishnan, a former engineer and venture capitalist who writes the Strange Loop Canon Substack, suggested that in awarding AI researchers twice the Royal Swedish Academy of Sciences is recognizing that the technology has already made them realize that powerful research breakthroughs can come from any field. “The committee is being more forward thinking in considering the end impact, and not just the precise qualifications or categories,” Krishnan said. “What this says specifically about AI is perhaps less surprising for those that think the technology is revolutionary … but what it says about our flexibility in going past the old paradigm is more interesting.” To people who watch the science prizes closely, there’s a telling difference between the two awards. While the connection between AI and the protein-mapping breakthrough is quite obvious, the connection between Hinton’s achievements and physics is less clear. Hinton, in his interview with the NYT, nodded to this: “If there was a Nobel Prize for computer science, our work would clearly be more appropriate for that. But there isn’t one.” “The two AI Nobels feel very different to me,” entrepreneur and AI critic Gary Marcus told DFD. “Certainly both Hopfield and Hinton have had considerable influence on AI, but the prize seems to represent some kind of shift in thinking in what the Physics prize is about.” That “shift in thinking” is what the two prizes for AI tools have revealed, even more so than those tools’ well-lauded achievements. The overwhelming power of pattern recognition boasted by modern AI models blurs the lines around who can discover what, in which field, in a way that makes the future of science vastly unpredictable. That might be a problem for 20th century institutions like the Nobels, or universities, or government agencies that use siloed categories to allocate funding and regulate — but these two prizes show that the people responsible for doing so should start thinking seriously about the problem.
| 
