"The Nobel, awarded to David Baker of the University of Washington and Demis Hassabis and John M. Jumper of Google DeepMind, is the second this week to involve artificial intelligence.
The Nobel Prize in Chemistry was awarded on Wednesday to three scientists for discoveries that show the potential of advanced technology, including artificial intelligence, to predict the shape of proteins, life’s chemical tools, and to invent new ones.
The laureates are: Demis Hassabis and John Jumper of Google DeepMind, who used A.I. to predict the structure of millions of proteins; and David Baker of the University of Washington, who used computer software to invent a new protein.
The impact of the work of this year’s laureates is “truly huge,” Johan Aqvist, a member of the Nobel Committee for Chemistry, said on Wednesday. “In order to understand how proteins work, you need to know what they look like, and that’s what this year’s laureates have done.”
That task once took months, or even decades. But A.I. models like AlphaFold make it possible to do that in a few hours or even minutes.
That speed has real-world applications. AlphaFold has been cited in scientific studies more than 20,000 times, and biochemists have used the technology to accelerate the discovery of medicines.
“We can draw a straight line from what we do to people being healthy,” Dr. Jumper said.
It could also lead to new biological tools such as enzymes that efficiently break down plastic bottles and convert them into materials that are easily reused and recycled.
Wednesday’s prize was the second this week to involve artificial intelligence, highlighting the technology’s growing significance in scientific research.
“A.I. is changing the way we do science,” said Frances Arnold, a chemical engineering professor at the California Institute of Technology who received the Nobel Prize in Chemistry in 2018. “It is supercharging our ability to explore previously intractable problems.”
Mary Carroll, the president of the American Chemical Society, said the selection of this year’s laureates indicated that the field of chemistry might be set to involve more computational study, which may improve the efficiency of chemistry research while making scientists less reliant on work in laboratories.
“Computational work is experimental work — it’s just a different kind,” she said. “I think it is a direction of chemistry.”
This year’s Nobel Prize in Chemistry also offered a reminder of how A.I. could be co-opted by bad actors.
“Of course it’s a dual-purpose technology,” Dr. Hassabis said at a news conference. “It has extraordinary potential for good, but also it can be used for harm.”
Some worry that this technology may be used to create new viruses or toxic substances that could be used in biological attacks. Dr. Baker was one of more than 90 scientists who signed an agreement this year that sought to regulate the equipment needed to manufacture new bioweapons, an effort to ensure that their A.I. research will not cause harm.
Cracking the Code
Proteins and enzymes are the microscopic mechanisms that drive the behavior of viruses, bacteria, the human body and all other living things. They begin as strings of chemical compounds, before twisting and folding into three-dimensional shapes that define what they can and cannot do. Pinpointing the precise shape of individual proteins was a laborious task for many years, and scientists had struggled for over 50 years to solve what was called “the protein folding problem.”
Demis Hassabis was born in London, where his parents — one a Greek Cypriot, the other a Singaporean — ran a toy store. As a teenager, he was the second-highest-ranked chess player under 14 in the world, and he began designing video games professionally before attending college.
After completing a computer science degree at the University of Cambridge, he founded a video game company, then returned to academia for a doctorate in neuroscience. He and a fellow academic, Shane Legg, and a childhood friend, Mustafa Suleyman, founded an A.I. start-up called DeepMind in 2010. About four years later, Google acquired it for $650 million.
DeepMind’s stated goal was to build artificial general intelligence, a machine that can do anything the human brain can do. It also explored other technologies that could help reach that goal and solve particular scientific problems. One of those technologies was AlphaFold.
AlphaFold is built using a mathematical system called a neural network. With neural networks, computers can analyze vast amounts of data to learn to perform many tasks that were once beyond their capacity. Such systems drive facial and voice recognition, as well as online chatbots.
They can also be used to predict the shape of a protein in the human body, which can determine how other molecules will bind or physically attach to it. This is one way drugs are developed: A drug binds to particular proteins in the body and alters their behavior.
John Jumper, the youngest chemistry laureate in over 70 years, was born in the United States. After finishing an undergraduate degree at Vanderbilt University and a master’s degree at the University of Cambridge, he earned a Ph.D. degree in theoretical chemistry at the University of Chicago.
He joined DeepMind as a researcher in 2017 after Google had acquired the lab. Alongside Dr. Hassabis and others, he soon began work on what would become AlphaFold.
In 2018, a DeepMind team led by Dr. Jumper entered a global competition called the Critical Assessment of Structure Prediction, a 25-year effort to solve the protein-folding problem. Their technology outperformed all other competitors.
Many scientists had assumed that a protein-folding breakthrough was still years away. Then in 2020, when the Google researchers unveiled an update of the technology, AlphaFold2, at the next contest, they showed that it had fully cracked the problem, predicting shapes with an accuracy level that rivaled physical experiments.
With AlphaFold2, the Google team was able to calculate the structure of all human proteins, the Nobel committee said, before eventually predicting “the structure of virtually all the 200 million proteins that researchers have so far discovered when mapping Earth’s organisms.”
Dr. Hassabis said he modeled DeepMind after Bell Labs, a research and development company that has produced 10 Nobel laureates. Researchers in fields like A.I. require a lot of resources and computing power, he said.
“Why not tap into the private sector in order to fund those kinds of things?” he said.
‘I Love All Proteins’
David Baker’s work preceded the emergence of the latest A.I. models and centered on protein creation.
A Seattle native, Dr. Baker earned his undergraduate degree from Harvard in 1984 and a doctorate in biochemistry doctorate from the University of California, Berkeley, in 1989. He now serves as the director of the Institute for Protein Design and a professor of biochemistry at the University of Washington.
In 2003, Dr. Baker and his colleagues created the first entirely new protein: a molecule called Top7 that was useless but symbolic [1].
“Until then, really the only proteins that were known were the ones that came down through millions or billions of years of evolution,” he said in an interview with The New York Times.
The researchers started with their desired protein shape and used a computer model called Rosetta, which searches databases of existing proteins to find a sequence of amino acids that might create such a structure.
He remembered the “amazing moment” when the protein he had created with bacteria from the proposed amino acid sequence showed almost the exact same structure as the one from his model.
This work “opened up a completely new world of protein structures that we had never seen before,” Dr. Aqvist of the Nobel committee said.
Dr. Baker realized that if he could create a novel protein structure, he should also be able to create more sophisticated proteins “that actually do things,” like break up the amyloid fibrils that are thought to be involved in Alzheimer’s disease.
In recent years, his work has dovetailed with the kind of research explored by Dr. Hassabis and Dr. Jumper at DeepMind, as his lab uses neural networks to not just predict the shapes of proteins but also generate blueprints for new proteins. It is another form of what researchers and tech companies call generative A.I.
His lab’s proteins — created with a more advanced iteration of Rosetta — have already been the basis of several potential medical treatments, like an antiviral nasal spray for Covid-19 and a medication for celiac disease. A Covid-19 vaccine, SKYCovione, based on his one of his lab’s proteins, was approved for use in South Korea in 2022.
Dr. Baker is also a co-founder of more than 20 biotechnology companies.
When asked by a journalist after the ceremony if he had a favorite protein, he said: “I love all proteins. I don’t want to pick favorites.” [2]
2. Nobel Prize in Chemistry Goes to 3 Scientists for Predicting and Creating Proteins. Moses, Claire; Metz, Cade; Rosenbluth, Teddy. New York Times (Online) New York Times Company. Oct 9, 2024.
Komentarų nėra:
Rašyti komentarą