How a gaming teenager defeated an academician and won the Nobel Prize

Source: Shihan

Today I want to tell you an extremely inspiring story in the field of science, a story that sounds incredible: a story about a young gamer who defeated a scientist to win the Nobel Prize.

Yes, I have talked about how gaming graphics cards triggered a computing revolution and then incubated the AI ​​industry before, but today's story is even more incredible and has more twists and turns than that.

There was such a young man who was hailed as a genius since he was a child and loved games since he was a child. When he was 4 years old, he showed a strong interest in chess. By the age of 8, he had won the championship in the official chess competition. He used the prize money to buy himself an important gift, a computer~ and soon fell in love with computer games.

Familiar operation, familiar plot~

At the age of 17, he chose to join a game company and become a game designer.

After all, since you love games so much, why not try to make one yourself? He joined the then famous Bullfrog Company.

One year after joining the company, he led the design of a hit game, the famous "Theme Park".

In short, this game in 1994 is the ancestor of many theme park and simulation management games today. I even think that the Tropico series should be influenced by it.

A few years later, he founded his own game company and successively developed two games, "Republic" and "Evil Genius", both of which are simulation management games.

Obviously, he likes the simulation management game type very much.

Civilization 5, start!

At this point, this story seems similar to the chess genius and computer prodigy stories mentioned in the past.

He loved games since he was a child, had a talent for chess and go, was able to teach himself computers, and finally joined a game company to become a super programmer, creating a hit product that was famous in the industry.

But the outrageous things about this guy have just begun.

After making the hit game, he soon began to think about the role of the computer as a tool for games, and he began to try to add AI functions to the game.

Few media have mentioned this, but as a veteran game player, I think this is likely the influence of the previous few games.

Because those who often play business simulation games can feel that in the late game, when there are a large number of NPCs, the computer computing power will have obvious shortcomings.

In the late game of Civilization 5, the computer often freezes in one round.

In the late game of Theme Park, Skyline, Tropico, etc., not only the graphics are stuck, but also the citizens' commuting routes are very unreasonable. Even if you build buses and subways for them, and even put residential areas and work areas together, they will run around and block the road.

It may be these phenomena that inspired him to think about AI. Can these gameplay problems be optimized by AI?

In 2010, he founded a new company with the goal of "solving intelligence problems" and trying to master games with learning algorithms.

In 2013, they created an algorithm called Deep Q-Network (DQN) that can play computer games at a level beyond humans.

The test results show that DQN became the best player in the game Space Invaders within 30 minutes of starting the game.

In 2016, the company released another game AI and defeated the original world champion of this game.

——This time the AI ​​is called AlphaGo.

Yes, strictly speaking, you can understand the sport of Go as a game and AlphaGo as a game AI.

It’s just that the game of Go is special because of the nearly infinite calculation variables. It was once considered unbreakable. AlphaGo is also much smarter than those simple computers and crazy computers~

Many people are shocked by the wave of development of artificial intelligence in the past two years, and often regard it as a single, sudden thing.

Actually, it is not. The birth of countless video games in recent years has spawned a huge demand for game AI. Many players hope to fight against smarter AI in the game, or fight side by side with smarter NPCs. These demands force programmers to continuously strengthen their exploration of AI algorithms.

There has never been a programmer who whimsically said, "I will study a smarter AI at all costs. No."

The reality is that if you design a good algorithm, your game will be more fun and you will make 100 million yuan. His game will be smarter and sell 200 million yuan. It is the heavy bonuses that give everyone endless enthusiasm to invest in AI development.

Gunpowder was not designed from the beginning. No scientist has ever said that he will invent gunpowder today. It does not exist. It is a bunch of alchemists who hope to live forever. In order to meet the demand for immortality, they tinker with alchemy every day, adding a little of this today and a little of that tomorrow. In the end, they found that sulfur, saltpeter and charcoal mixed together will explode.

Leeuwenhoek didn't want to discover the microbial world at first. He was just a lens maker, polishing lenses every day. One day, he suddenly found that after polishing the lenses to the extreme, he could see things that the naked eye could not see.

The protagonist of our story is the same. At first, he wanted to make games, then he wanted to study smarter games, and finally he developed an extremely intelligent game AI.

Then they suddenly began to think about a question.

Since AI has the ability to self-learn and can quickly master the rules of Go and video games to become a champion player,

If we also understand scientific research in a certain field as a "game", can AI master it?

In 2017, at the Wuzhen Go Summit, AlphaGo defeated the world Go champion Ke Jie with a score of 3:0.

In 2018, DeepMind tried to develop an AI system that can predict protein structure, AlphaFold. Try to use AI for scientific research.

You must think this is unreliable. It is too far-fetched to let an AI originally designed for games study science.

You are not the only one who thinks so. A certain academician of the Chinese Academy of Sciences also thinks so.

Yes, it is our old acquaintance, Professor Yan Ning.

So all encounters in the world are reunions after a long separation. We actually met again after a twist of fate~

For many years, there are three main ways to predict protein structure. One is to use X-rays to illuminate protein crystals, the second is nuclear magnetic resonance spectroscopy, and the third is expensive cryo-electron microscopy photography and modeling.

Yan Ning's team is famous for its proficient operation of the third method, cryo-electron microscopy. Her team can take five photos in the time it takes others to take one photo, which is much more efficient.

And DeepMind's idea is, can this highly repetitive work be solved by AI?

If we understand the process of cryo-electron microscopy photography and modeling as a game, can we use AI to try to solve it?

"They didn't plan to take a film, but chose AI: since proteins are made up of amino acids, they only need to use the known protein structures that are publicly available everywhere, summarize the distances and connection angles of each pair of amino acids in these proteins into a picture, and then use the neural network to digest them. AI can make predictions by itself."

The final result is that AI is much more efficient than manual labor. The efficiency of a general team is 1, the efficiency of Yan Ning's team is 5, and the efficiency of AI is 100,000, and it is still growing rapidly. Because AI does not need to rest and will continue to evolve. Since their breakthrough, more than 2 million people from 190 countries have used AlphaFold. With their help, scientists can not only gain a deeper understanding of antibiotic resistance, but also design enzyme proteins that can digest plastic.

Such a subversive result, you should have guessed the story behind it. This technology won the Nobel Prize. This guy who loves games and originally worked as a game designer is this year's Nobel Prize winner in Chemistry, Hassabis.

Facts have proved that the development of the times will fairly throw everyone away. When you are stunned by the development of AI, top scientists may also make mistakes.

When we discussed AI in 2022, the impact of AI on Yan Ning and others has been observed by many people. Judging from the comments, although everyone recognizes the development of AI, most people think that it may take some time to replace top scientists. (Several friends spoke very forward-looking, very impressive)

Yan Ning herself may think so too. In 2022, Yan Ning's own conclusion is that AI's prediction level can only reach their 2017 level.

This plot is exactly the same as the Go industry.

When AlphaGo came out at the beginning, everyone thought it was nothing, it could only beat the world champion, and humans had a chance to win back with hard work.

But soon everyone found that this view was ridiculously wrong, because humans learn from teachers and textbooks, and human combat effectiveness is actually based on the experience of predecessors, plus the result of many years of learning, and AI has been in contact with Go for less than a year~one year after getting started, and you can beat the Go master, and you don't need to watch it in the future.

In 2022, Yan Ning felt that AI would only reach the level they had five years ago, so there was no need to worry.

The problem is that AlphaFold was launched in 2018, and it will only be four years until 2022. A four-year-old child is about to catch up with your top human scientists. If you use common sense to judge this development speed, you will definitely be wrong.

So what does this story tell us?

Is it the development of science and technology, the innovation of AI, the fate of life, or should biology turn into code farmers?

I think the biggest inspiration lies in passion.

Looking back, in 2007, Yan Ning was already a professor and doctoral supervisor at Tsinghua University, and a well-known academic master.

At this time, Demis Hassabis was still a game designer, let alone an academic master, he could not even be considered a member of the academic community.

At this time, if you tell him that you will beat the academicians and win the Nobel Prize in the future, he will not even believe it, let alone imagine it.

It is incredible that an unknown scientist won the Nobel Prize. But it makes sense.

How can I, a lousy gamer, win the Nobel Prize? There is no Nobel Prize for games, right?

This is the wonder of the world.

You don’t really love scientific research. Maybe it’s for the salary, maybe for stability, maybe for the bright lights. You do similar work day after day and feel deeply that scientific research is not easy.

Although he only made games, he loved games from the bottom of his heart. As a result, he studied them to the extreme and actually pointed out the AI ​​technology tree, which turned out to be the key to the new era.

You say he was lucky, but if he didn’t have an extreme love for games, if he didn’t think about the gameplay of games from the root, if he just made some skin-changing games to make money, would this story happen? Obviously not.

It was the love and study of things that surpassed everything that helped him to break through the fog and find a new world.

Never forget to love what you love.