On October 9th local time, the Royal Swedish Academy of Sciences announced that the 2024 Nobel Prize in Chemistry was awarded to David Baker, Demis Hassabis, and John M.Jumper for their contributions to protein design and protein structure prediction. The three laureates shared a prize of 11 million Swedish kronor (approximately 7.45 million yuan). The Nobel Prize committee commented that Baker, from the University of Washington in the United States, successfully accomplished the nearly impossible task of constructing entirely new proteins; while the British scientists Hassabis and Jumper from Google developed an artificial intelligence model named AlphaFold2, which solved a 50-year-old problem by predicting the complex structures of about 200 million known proteins and has been used by more than 2 million people worldwide.

One of the awardees, Demis Hassabis, was once described by the great British physicist Stephen Hawking as "one of the smartest humans on Earth." He is also the founder of the British artificial intelligence company DeepMind. In 2014, DeepMind was acquired by Google for $600 million, even though the company had not yet publicly released any products and only had 20 technical staff. However, Hassabis and his team have since influenced Google's development direction for the next decade, prompting the tech giant to shift from mobile-first to AI-first.

Advertisement

Before OpenAI, the last global wave of artificial intelligence enthusiasm was sparked by DeepMind's AlphaGo. Hassabis was affectionately referred to as the "father of AlphaGo." A few years later, DeepMind's AlphaFold made outstanding contributions in the field of protein prediction.

In May 2017, this AI star with 1/4 Chinese heritage came to China for the first time due to AlphaGo's match with Chinese Go player Ke Jie and accepted an exclusive interview with First Financial Daily after the match. The business card he handed to the reporter from First Financial Daily had his Chinese name printed on it.

At the time, in his 40s, he was somewhat balding and resembled a typical teacher at a Chinese university. However, when he spoke of his mission to "make artificial intelligence the ultimate tool for exploring the universe," no one doubted his sincerity.

During his interview with First Financial Daily, he said that in the field of artificial intelligence, the two things that excited him the most were deep learning and reinforcement learning. AlphaGo was the combination of the two and an important step towards the goal of general artificial intelligence. Hassabis said, "First, solve intelligence, then use intelligence to solve everything."

From a Child Prodigy to a Leader in AI

Hassabis can be considered a "child prodigy." He enjoyed various intellectual games and started playing chess at the age of 4. By the age of 13, he had obtained the title of chess master and still holds the world record for winning the "Intellectual Olympics" Elite Championship five times.

At the age of 11, participating in an international circuit in Liechtenstein left a deep impression on young Hassabis.At the time, he was playing a match against a Danish chess champion. After 10 hours, Hassabis was left with only his king and queen, while his opponent had a king, a rook, and a bishop, and was in a dominant position. The game could have ended in a draw, but Hassabis was too exhausted and chose to resign.

Hassabis recalled in a BBC program that this match led him to an epiphany—"Are we wasting our brains? Since the players at this level are all top-notch, why not use our mental power to do something more meaningful, such as solving cancer, finding cures for other diseases, wouldn't that be better?"

At that time, Hassabis ranked second in the world among chess players of his age, and almost everyone around him took it for granted that this would be his future career. However, Hassabis did not embark on the path of a professional chess player, but instead developed a strong interest in computers.

In school, several friends and Hassabis formed a hacking club, writing code together, creating image presentations, copying movies they had seen or games they had played, and spending all their spare time on how to use these computers for better programming.

At the age of 20, Hassabis obtained a degree in computer science from the University of Cambridge and later founded the video game company Elixir. He commented that successful games reflect life from a certain perspective and can teach us a lot, "Life does not allow us to always pursue stability; it needs to constantly stimulate us to generate new ideas, improve strategies, and develop our own brains, and games are like brain exercises."

Afterward, Hassabis chose to return to academia, pursuing a Ph.D. in neuroscience at University College London, completing academic research on the hippocampus and episodic memory, and in 2011, co-founded the DeepMind team with artificial intelligence expert colleague Shane Legg and serial entrepreneur Mustafa Suleyman.

As a genius "unlike others," Hassabis has his own unique "biological clock": he starts work at 10 a.m., spends the whole day in the DeepMind office, goes back to have dinner with his family, and then starts working on the next day at 10 p.m., not going to sleep until 4 a.m. Usually, he spends several hours at night conducting research, reading the latest academic papers, and engaging in creative thinking.

In 2014, Google acquired DeepMind for $600 million, which was the largest acquisition Google had made in Europe at the time, and DeepMind had not yet publicly released any products, with only 20 technical staff. These rich experiences may not be unrelated to Hassabis's family education. "My background is very diverse," Hassabis said in an exclusive interview with First Financial Daily, his father has Greek and Cypriot ancestry, and was a creative singer in his youth, while his mother is a Singaporean Chinese, and they both like to do things according to their preferences, which influenced him to—"Don't follow the beaten track, go your own way, and keep going, that's the right way to live."

With Google's "blessing," DeepMind accelerated the development of AlphaGo. Until AlphaGo defeated the once world number one Go player Lee Sedol with a score of 4:1, it was considered a historic breakthrough in the history of artificial intelligence development.

However, just after AlphaGo was newly crowned the Chinese Go 9-dan, Hassabis announced that it would withdraw from the competitive stage, and the AlphaGo development team would focus on other major challenges, developing advanced general algorithms to help scientists solve the most complex problems, including finding new disease treatments, significantly reducing energy consumption, and inventing revolutionary new materials.In addition to managing DeepMind, Hassabis has also published numerous papers at top-tier global scientific conferences. In April last year, Google announced the merger of Google Brain and DeepMind to form a new division called Google DeepMind. The new division aims to lead groundbreaking AI product research and progress while maintaining ethical standards. Soon after, the two departments joined forces to create Gemini, a ChatGPT counterpart, by the end of 2023.

By May of this year, AlphaFold 3, developed by the DeepMind and Isomorphic Labs research teams, made it to the cover of Nature magazine. AlphaFold 3 successfully predicted the structures and interactions of all molecules of life, including proteins, DNA, RNA, ligands, etc., with unprecedented accuracy.

In the latest interview in August this year, Hassabis also expressed his latest views on AI: In the short term, AI is overhyped, but in the long term, it is underestimated. As for how to distinguish, Hassabis said, "In addition to doing research, one must also look at the background of the person making the statement, how technologically knowledgeable they are, and whether they only switched to AI from another direction last year. If the person making the statement is just following the trend, the probability of them contributing good ideas will be as likely as a lottery draw."

He also predicted that AI might make breakthroughs in some complex mathematical problems, such as helping to solve famous mathematical conjectures or performing well in international mathematics competitions. An important test criterion for AGI will be whether it can independently generate new hypotheses and theories like general relativity.

Exploring the "cosmic ocean" of AI

At a TED conference, then-Google CEO Larry Page spoke at length about Hassabis and referred to DeepMind as "one of the most exciting things I've seen in a long time."

The success that Hassabis has led DeepMind to achieve in the field of artificial intelligence has clearly convinced Google that it is time to make researching artificial intelligence a top priority for the company.

"We are very pleased to prove that the cutting-edge technology is not only present in Silicon Valley. I am proud to be British and to carry the banner of British innovation," Hassabis told First Financial.

In his view, the UK has always had an innovative tradition in researching computer technology and artificial intelligence, such as Alan Turing and the "father of the internet" Tim Berners-Lee. It's just that, perhaps, everyone is not so good at transforming technology into commercial success."Many places can conduct cutting-edge research, just like London and China, why not? As long as you have enough smart people and give them a chance," said Hassabis.

Standing behind Hassabis and DeepMind are some of the brightest talents from over 60 countries, continuously solving the most interesting challenges in the field of artificial intelligence. Hassabis said with a smile: "The top talents always want to work with other top talents in the world." Especially when you start to create products like AlphaGo or publish research results in Nature magazine, it is an advertisement to the best talents in the world, "If they want to create great products like AlphaGo, then DeepMind is a good place to work."

On the other hand, these top talents often want to study the most interesting questions and accept the most interesting and intelligent challenges. Hassabis believes that solving artificial intelligence problems is one of the most interesting things you can do, so for those extremely smart people, the challenges encountered in research are also very interesting.

At that time, the reporter asked him how far AlphaGo is from strong artificial intelligence, and even super artificial intelligence?

Hassabis replied that it is like some super artificial intelligence in science fiction movies, we are still far from that goal, at least there are still several decades to go, "Because there are some key issues in the field of artificial intelligence that we have not yet conquered."

"If artificial intelligence can discover new knowledge and strategies in the aforementioned fields, then the breakthroughs in these fields will be very considerable. We can't wait to see all this happen," said DeepMind executives, including Hassabis, in a blog.

At that time, Hassabis gave an example to the reporter, through cooperation with human experts, a variety of innovative methods can be found, "including applying one variant of AlphaGo to the medical industry, which is focusing on solving the problem of protein folding, to treat Alzheimer's disease."

Unexpectedly, seven years later, the "solving the problem of protein folding" mentioned by Hassabis in the interview at that time, due to a major breakthrough, won the 2024 Nobel Prize in Chemistry.

The human-computer war has long ended, and the era of large models has begun. But everything is just as Hassabis said in an interview many years ago, "Human wisdom will be amplified by artificial intelligence. Artificial intelligence and AlphaGo are tools, just like the Hubble Telescope, which can promote the progress of human civilization."