Two Nobel Prizes for AI, and a warning we should not forget

In barely two days, the Royal Swedish Academy of Sciences has done something that would have seemed unlikely ten years ago: awarded artificial intelligence in two categories.

On 8 October, the Nobel Prize in Physics went to John J. Hopfield and Geoffrey Hinton “for foundational discoveries and inventions that enable machine learning with artificial neural networks”. The following day, the Chemistry prize was shared between David Baker, for computational protein design, and Demis Hassabis and John Jumper of Google DeepMind, for protein structure prediction with AlphaFold, a problem that had remained open for some fifty years.

From tool to science

For those of us who apply AI in businesses, these prizes have a practical reading. For years, machine learning was seen as an engineering technique: useful, but somewhat opaque and not quite “serious”. The Nobel gives it a different standing. Neural networks are no longer just what sits behind a series recommender; they are a tool that produces new scientific knowledge.

AlphaFold is the best example. It doesn’t replace biologists; it saves them years of experimental work so they can devote their time to what only they can do. That is, on a small scale, the same promise we look for in any business process: take the mechanical part off people’s plates so they can focus on the part that requires judgement.

Hinton’s warning

But the most striking detail of the week is not in the official press releases. Geoffrey Hinton left Google in 2023 precisely so that he could speak freely about the risks of AI. On the day of the announcement, according to The Guardian, he compared its impact to the Industrial Revolution and added that he worries the ultimate consequence may be systems more intelligent than us that end up taking control.

You may or may not agree with his diagnosis of the long term. I don’t have a settled view. But I think it healthy that one of the people who has done most for this technology is using the biggest megaphone in the scientific world to call for research into its safety as well.

What I take into everyday work

Between enthusiasm and fear there is a reasonable middle ground, and that is the one that interests me:

• Use AI where it genuinely adds value, just as AlphaFold adds value where the traditional method was painfully slow.
• Keep a person accountable for every result that matters, however good the model.
• Take those who sound the alarm seriously, even when we disagree, especially when the person sounding it is the one who knows the system best.

It is a good week for AI. It is also a good week to remember that the best science comes with doubts attached.