Nobel Prizes for Physics and Chemistry 2024: landmarking the age of AI
By Emer McGovern
From chatbots to Garmin watches, AI has already transformed how we live, with some scholars predicting its paradigm-shifting impacts to have a historical status emulating the Industrial Revolution.
This has been demonstrated most recently in awarding this year’s Physics and Chemistry Nobel Prizes to AI-related discoveries. John J. Hopfield and Geoffery Hinton were awarded the Physics Nobel for their work in developing neural networks, which form the basis of machine learning that underpins modern AI. The Chemistry Nobel Prize was awarded to the creators of AlphaFol, an AI program credited with solving the enigmatic ‘protein folding problem’.
In 1982, Hopfield, a theoretical biologist, unveiled a simple dynamic network to explain memory storage in the brain. Hinton later built on this using statistical physics to develop the Boltzmann machine, a neural network that could generate new patterns in addition to storing them. Although both scientists came from a biological perspective, the development of this system established a basis for the retention of information in machine learning and generative algorithms that can store and synthesise information, such as the Natural Language Processing algorithm used in ChatGPT.
Protein structure is integral to its function. How a linear sequence of amino acids (subunits that make up proteins) can result in spontaneous folding into highly specific structures has puzzled scientists for decades. From the mid to late 20th century, Nobel prize winner Christian Anfinsen and molecular biologist Cyrus Levinthal demonstrated that the amino acid sequence dictated the folding and, thus, structure of proteins through local interactions in the amino acid chain. This launched the ‘protein-folding problem’: how can we decode and thus predict protein structure based on amino acid sequence? With proteins having pivotal roles in disease, to do so would revolutionise drug design. David Baker, Demis Hassabis, and John Jumper have been accolades for doing just that, from Baker’s foundational work on short amino acid sequences to Hassabis and Jumper’s collaboration on developing AlphaFold, a computational model that predicts protein structure from amino acid sequence inputs.
AI is rapidly becoming a defining feature of our era. Some, including the prize winners, see this as a warning sign that our world is being transformed by AI at such a fast pace that there is a risk of falling behind. The fear of technology surpassing and outwitting humanity has been depicted in numerous sci-fi films, but Physics Nobel Prize winner and “godfather of AI” Geoffrey Hinton believes it to be a real possibility beyond the realm of cinema.
Following a long line of ‘Frankenstein-esque’ scientists who have grappled with the realities of their inventions, Hinton has expressed regret about his discoveries and has been outspoken about the risks of uncontrolled AI development. Talking to 60 Minutes in 2023, Hinton commented that ‘’(he) can’t see a path that guarantees safety’’ given we’re ‘’moving into a period when for the first time ever we may have things more intelligent than us,’’ and fears properties inherent to the neural network framework, self-learning and ability to share code across models facilitate AI to ‘’escape control’’ and ‘’be able to manipulate people.’’
“Moving into a period when for the first time ever we may have things more intelligent than us”
Geoffery Hinton
This sentiment has been mirrored among other AI experts with an open letter signed by Elon Musk, Apple co-founder Steve Wozniak, and researchers at DeepMind in 2023 to ‘’immediately pause for at least six months the training of AI systems more powerful than ChatGPT-4.’’Despite claiming in mid-2023 that OpenAI, the founding company, had no intentions to train a successor to ChatGPT-4, ChatGPT-5 is expected to be released soon, with several people anticipating the launch to be announced at the developer day conference in November this year.
The company has so far been cryptic about the exact advancements of the new model, but it has been revealed that ChatGPT-5 has been trained on larger and more diverse data sets, improving accuracy and comprehension. We can expect enhanced multimodal capabilities, allowing the input of videos in addition to images and text.
As Hinton warned us, the race to develop more intelligent and useful AI for consumers, businesses, and academia alike has resulted in rapid technological changes and shifts in our relationship with it. How can we prevent a dystopian science fiction future where humanity is in conflict with technology? We need to embrace AI in its entirety: the challenges it poses to our safety and the ways it can make our daily lives easier. At this point, the question is not whether we will use AI but how we can use it in a way that helps us more than it harms us.
Image Niklas Elmehed via Nobel Prize Outreach

