NOBEL PRIZES 2024: Physics and Chemistry
Photo by Alexander Mahmoud, 2018. Art design: Olena Burdeina (FA_Photo) via Photoshop
This year, the Nobel Prizes come with quite a few «oddities». The discoveries for which the Chemistry Prize was awarded would have seemed more fitting in the category of scientific achievements in physiology and medicine.
The breakthroughs that enabled machine learning relate more to mathematics and computer science than to physics. Physics was the first field of science mentioned in Alfred Nobel’s will, but as is well known, no prize is awarded for mathematics. So, where do these «oddities» come from?
It appears that the 2024 prizes have captured a trend: AI has become irreversibly integrated into modern science.
NOBEL PRIZE IN CHEMISTRY: THE ANSWER WAS SOUGHT FOR 50 YEARS!
HOW TO PREDICT PROTEIN STRUCTURE
P
roteins, which control the body’s chemical reactions, are the most crucial tool nature uses to create diversity in life. Proteins form hormones and antibodies, signaling molecules, muscle, keratin, and other tissues.
The synthesis of proteins involves several dozen amino acids that coil into three-dimensional strands — unique combinations that determine the specific function of each protein. The sequence of amino acids in the protein entirely dictates this three-dimensional structure. But can we predict the structure based solely on this sequence? If we succeed, we could design proteins for any desired function!
Work in this direction has been ongoing since the 1970s but without much success. The problem lies in the fact that a protein can potentially adopt up to 10⁴⁷ forms. In a cell, this folding happens instantly. However, if the amino acid chain were to form randomly, finding the correct structure would take more time than the age of the universe.
WE CAN CREATE ANY PROTEIN!
This is where artificial intelligence came to the aid of scientists, capable of detecting patterns in vast amounts of data. After 50 years of fruitless scientific searches, Demis Hassabis and John Jumper’s development of an AI model to solve this problem was met with extraordinary enthusiasm by the scientific community.
The scientists created AlphaFold2 — an AI model capable of predicting the structure of 200 million proteins found on our planet! It has now become possible to design any protein. However, Hassabis and Jumper are not the only Chemistry laureates this year. The prize was also awarded for another discovery.
The 11 million Swedish kronor prize (just under $1.1 million) will be shared among three. One half will be split between Jumper and Hassabis, and the other will go to David Baker from the Howard Hughes Medical Institute for «computer-based protein design».
AI DESIGNS PROTEINS
David Baker approached the problem «from the other side» and managed to understand how exactly a protein forms its unique structure. His software, called Rosetta, was not developed to predict the structure of a protein but to directly define the desired model, after which Rosetta would generate the corresponding amino acid sequence.
This gave us the ability to create proteins with new functions that do not exist in nature. The first such protein, Top7, was successfully designed and produced in laboratory conditions. The laureates have made these tools openly accessible, and now, more than 2 million scientists from 190 countries are using AlphaFold2.
A CHESS PLAYER AND GAME CREATOR
The laureates themselves are extraordinary individuals, especially Demis Hassabis, a British scientist and member of the Royal Academy of Engineering and the Royal Society. His father is Greek, and his mother is Chinese. From a young age, Demis showed exceptional talent in chess, ranking 5th in the world youth rankings.
After graduating from Cambridge with honors, he became fascinated with AI systems for games. He contributed to the development of video games such as Black & White, Republic: The Revolution, and Evil Genius. After earning a Ph.D. in neuroscience, Hassabis founded DeepMind in 2010, which was acquired by Google in 2014 for £400 million.
Incidentally, the second laureate, 39-year-old John Jumper, is also a DeepMind employee. The company is widely known for developing self-learning neural networks like AlphaGo, AlphaZero, and AlphaStar. DeepMind gained worldwide fame when AlphaGo defeated a world champion in the game of Go, and for creating a neural network capable of playing video games like a human.
NOBEL PRIZE IN PHYSICS: THE «GODFATHER OF ARTIFICIAL INTELLIGENCE» FEARS HIS CREATION
NO ONE EXPECTED THEM TO WIN THE PRIZE
Two scientists — Professor John J. Hopfield from Princeton University and Professor Geoffrey E. Hinton from the University of Toronto — received the Nobel Prize in Physics for «developing methods that became the foundation of modern machine learning».
«John Hopfield created an associative neural network capable of storing and retrieving images and other types of data patterns. Geoffrey Hinton invented a method that autonomously discovers properties in data and performs tasks such as identifying specific elements in images».
This was how the Royal Swedish Academy of Sciences justified its decision. However, in scientific circles, the choice caused some bewilderment. Typically, publications dedicated to physics issues predict Nobel nominees years in advance.
But these laureates had never been mentioned in those forecasts. Thus, the Nobel Committee’s surprise was successful — no one expected that «outstanding physicists» would be announced as specialists in computer science and neurobiology.
A NEW METHOD OF KNOWLEDGE
It can be assumed that the academics highlighted the immense importance of neural networks for science in this way. Today, no serious research in physics can be imagined without machine learning algorithms.
Although Hopfield’s and Hinton’s discoveries were not made in the field of physics, they have significantly expanded the capabilities of physical science. They introduced the world to a new method of knowledge, allowing us to delve even deeper into the mysteries of nature.
The world is changing rapidly, and with it, the principles of awarding the Nobel Prizes are evolving. Alfred Nobel intended for the prizes to be awarded for discoveries made in the previous year, and computer science was not on that list — it did not exist in Nobel’s time.
In the case of Hopfield and Hinton, the breach of Nobel’s will is clear. However, the Nobel Committee’s reasoning is understandable: Nobel envisioned lifetime awards, but how should the contributions of scientists who made groundbreaking discoveries in the last century, yet are still alive, be recognized?
AND YET, THIS IS STILL PHYSICS
While this may contradict Nobel’s original principles, it does not go against the gratitude we owe to these scientists. Some believe the committee made this compromise because it’s increasingly difficult to find contemporary works in physics that match the great discoveries of the past.
However, Hopfield, who invented the associative neural network, and Hinton, a pioneer of the «backpropagation» method for training multilayer neural networks, have had a profound impact on science. Their ideas are still connected to physics, as they applied physical tools in constructing neural networks.
For example, in creating «Hopfield networks», the scientist used principles of quantum mechanics — system characteristics determined by atomic spin. Once revolutionary, the scientific achievements of these laureates are now used in diverse fields: climate modeling, solar panel development, medical image analysis, and much more.
BAD PEOPLE AND TERRIBLE THINGS
Many are now listening closely to Geoffrey Hinton, deservedly called the «Godfather of Artificial Intelligence». After leaving Google, where he had worked in recent years, the scientist has increasingly warned the world about the dangers that his creation may pose.
This sense of responsibility once weighed heavily on nuclear physicists, the creators of the atomic bomb. Defending himself, the 78-year-old scientist says, «I comfort myself with the banal excuse: if I hadn’t done it, someone else would have». Hinton is most concerned about the rapid pace at which AI technologies are advancing.
«What worries me is that this could lead to systems that are more intelligent than we are, which will ultimately take control», says the scientist. He fears that these systems will far exceed human capabilities, and then they may begin to control us.
Moreover, Hinton is frightened that «bad people capable of terrible things» might take advantage of this.
When copying materials, please place an active link to www.huxley.media
Select the text and press Ctrl + Enter
