Opinion
Yes, Nobel prizes for the work of AI pioneers deserve standing ovations
Summary
- A big moment for AI was its 1955 coinage, but this year’s Nobel haul qualifies too. Laureate Geoffrey Hinton, famously ‘a man who never sits down,’ had computers mimic the human brain for ‘deep learning’ while Demis Hassabis set up DeepMind, whose use of AI might well revolutionize healthcare.
The term ‘artificial intelligence’ was coined in 1955, when the original ‘fathers of AI,’ Marvin Minsky, Claude Shannon and John McCarthy, convened the Dartmouth Conference in 1956. Another seminal moment came in 2017, when the paper, ‘Attention Is All You Need,’ gave birth to the transformer and Generative AI.
There have been other big moments, but the technology came into its own in 2024, when its leading lights won two Nobel prizes for their work in AI. John Hopfield and Geoffrey Hinton won the Physics prize for their work in deep learning and associated fields.
The next day David Baker, Demis Hassabis and John Jumper won the chemistry prize for applying deep learning and AI to solve intractable problems of protein folding.
Two of these names are leading lights of AI: Hassabis, who set up DeepMind, and Hinton, a ‘father of AI’ who is probably the only person to win both the Turing Award and a Nobel.
Also read: Artificial intelligence and what it owes a man who never sits down
I have written often about them and count both as inspirations. It gratifies me even more that both are alumni of my own alma mater, Cambridge University.
Geoff Hinton from King’s College is also famously someone who does not sit down; his severe back problems prevent him from doing so. I quoted him in an earlier column as saying: “I last sat down in 2005, and it was a mistake."
With his wry British humour, he talks of his back being “a long-standing problem." His fame, however, derives from his storied achievements in AI: he rescued neural networks from an AI winter, ‘invented’ deep learning, and tutored a clutch of AI stars, including Ilya Sutskever.
After studying at Cambridge and Edinburgh, Hinton went to study something different: How the human brain stored memories, and how it worked.
He was one of the first researchers who started working on ‘mimicking’ the human brain with computer hardware and software, thus constructing a newer and ‘purer’ form of AI, which we now call ‘deep learning.’
His PhD thesis showed how deep neural networks outclassed older machine learning (ML) models at identifying speech patterns. He invented ‘backpropagation,’ which was one of the concepts that inspired Google’s BackRub search algorithm.
By mimicking the brain, he sought to get rid of traditional ML techniques, where humans would label pictures, words, and objects; instead, his work copied the brain’s self-learning techniques.
Also read: A Nobel for Google DeepMind is a tribute to how AI can transform drug discovery
He and his team built ‘artificial neurons’ modelled on the columns of neurons in the brain’s cortex. These neural nets can gather information, react to it and build an understanding of what something looks or sounds like.
The AI community did not trust this novel approach; Hinton told Sky News that it was “an idea that almost no one on Earth believed in at that point." Today, deep learning is behind the large language models (LLMs) of GenAI. It drives autonomous vehicles, predicts the weather and predicts how proteins fold.
Demis Hassabis graduated from Queens College at Cambridge much later in 2014. He took Hinton’s legacy further and created DeepMind, later bought by Google.
His lab was the original OpenAI in its singular pursuit of ‘solving intelligence’ and artificial general intelligence (AGI). DeepMind got famous when its AlphaGo defeated the Go world champion in 2017, but its real breakthrough was AlphaFold.
This AI product cracked one of the hardest problems in medical science: predicting how a protein would fold. Every carbon-based life form is made of proteins, and how they fold decides almost everything about our physiology and life.
There are over 200 million known proteins today, and each of them folds in a unique 3D shape. It was impossible for scientists to study each one of them. If proteins fold wrongly, for instance, they can cause horrific harm; the accumulation of misshapen proteins is said to cause Alzheimer’s, Parkinson’s, etc.
DeepMind took AlphaFold to a point where it can predict the folded shape of a protein right down to its molecular level. In 2020, the big problem was declared solved.
It was a breakthrough as important as mapping the human genome, or the discovery of antibiotics, something that can change medical science forever. As a scientist remarked, “What took us months and years to do, AlphaFold was able to do in a weekend."
Also read: Nobel for a prosperity theory: Let’s talk about it in the context of AI
Now LLMs and Generative AI are doing equally magical stuff. But it was the yeoman work done against all odds by pioneers like Hinton to take deep learning mainstream that gave AI the world’s biggest prize for intellectual pursuits.
Indeed, the Nobel awards were more for deep learning than AI. For a field pioneered by the restless man who never sits down.
