Unfortunately, the development of artificial intelligence technology for people on earth has taken a wrong path

​Planet Kepler-186F Cygnus.

Observer No. 9527 is observing the Earth boredly.

The director of the Universe Civilization Monitoring Bureau suddenly appeared.

9527, you have monitored the earth for a hundred years. How far has their so-called artificial intelligence developed?

They spent a lot of time playing chess.

Play chess?

Yes, it is two people sitting in front of the table, each holding something called chess pieces, and fighting on something called a chessboard.

fighting?

Well, it’s actually a game, to determine the winner or loser. In the words of earthlings, games are the source of knowledge, and they develop artificial intelligence by letting machines play chess.

The thinking of the people on Earth is really strange, so have they succeeded?

It seems to be successful, but it seems to be unsuccessful. If you have time, I can tell you in detail?

Don’t be verbose and focus on the key points.

People on earth especially like to deceive. As early as 1769, this is their AD calendar, a German made a mechanical chess-playing device, which was named "Turk" ", because behind the installation there is always a wooden man in Turkish attire. By the way, Turkey is another country to them.

​

Mechanical device that can play chess automatically? Is the technology tree of the people on earth growing crooked?

This "Turk" became a repertoire for entertaining European royal families, and was also dedicated to a military genius named Napoleon, the conqueror of Europe. Napoleon played chess with the Turk and came back defeated. He was so angry that he dropped all his chess pieces to the ground.

The "Turk" has been fooling around in Europe for decades. Later, it was discovered that it didn't know how to play chess at all. Instead, there was a living person inside who played chess. Master!

These strange earthlings, this is absolutely impossible to happen on our planet.

In the 1940s, for some unknown reason, a large number of geniuses suddenly emerged among the people on earth. They not only invented computers, but also began to think about such profound issues as whether machines can possess human intelligence. .

genius? Anyone?

For example, Turing, who laid down the basic theory of programmable computers on earth, von Neumann proposed a computing architecture that is still in use today, and Shannon, who actually discovered The secret behind information established information theory. Fortunately, he loved juggling and turned to "juggling" in the second half of his life. Otherwise, God knows what else he would have come up with!

People on earth are indeed very strange. Such a person on our planet has to drain his intelligence. Let him rest.

These geniuses on earth all like to study computers to play chess. Turing is the most pitiful. Although he wrote a chess program in 1947, the computer time at that time (referred to as machine time) It's so precious that it won't be his turn at all.

Von Neumann proposed a two-person game algorithm called Minimax. The game process constitutes a game tree. The growth of the tree is exponential. When the tree is very deep, , the size of the tree will become uncontrollable.

​

So another genius McCarthy invented an optimization called α-β pruning Method: The original Minimax algorithm statically calculates the evaluation function after all the game trees are drawn, while α-β pruning adopts a dynamic method of calculating the evaluation function while drawing the tree. When the value of the evaluation function exceeds the given upper and lower bounds, the tree search process stops, thus greatly reducing the size of the tree.

Doesn’t that juggling Shannon play chess?

Definitely, this person seems to be particularly good at explaining things from a mathematical perspective. This is true for information theory and the same is true for chess. He wrote a paper "Programming to Realize Computer Chess", defining the chessboard as a two-dimensional array. Each chess piece has a corresponding subroutine to calculate all possible moves of the chess piece. Finally, there is an evaluation function, so that from mathematics Angle gives a method to evaluate the complexity of board games.

Then can their computers play chess?

I know how to play, but my level is very low, and I can't beat the people on earth at all, so humans have to let the machines play against each other, and hold some kind of "World Computer Chess Championship" to see who can win. The machine level is high.

Hahaha, poor earthlings!

At that time, there were two big countries on the earth, the United States and the Soviet Union. They were fiercely competing in various fields such as military, economy, technology, and sports. Machine chess was no exception. The one who represented the United States was Kotok-McCarthy, developed by McCarthy at Stanford. . After four months of fierce fighting, Kotok-McCarthy still lost to the Soviet program 1:3.

You just said that it took 4 Earth months to play four games of chess?

Yes, people on earth did not have the Internet at that time, and they used something called a telegraph to play chess. This was a communication tool with backward machines.

People on earth developed hundreds of thousands of years later than us. It is really unimaginable to have such slow communication efficiency.

Later, the United States developed a program called CHESS, which was invincible in the United States and won 4 computer chess competitions. The Americans were full of confidence and participated in the World Computer Chess Championship to take revenge on the Soviet Union. Unexpectedly, they were defeated by the Soviet KAISSA chess program. I have to say that the Soviet Union was very powerful in the field of chess, with great players and great programs.

Then the United States just gives up?

No, in the 1980s, another genius from Bell Labs in the United States, Ken Thomson, who invented Unix, developed the Belle program. It was very powerful and he was the first to obtain the "Master" "Computer chess player with the title.

When Belle went to compete in the Soviet Union and prepared to take revenge, she was unfortunately confiscated by the United States government at the customs at JFK Airport. It was an attempt to deliver advanced weapons to the Soviet Union because they contained very large-scale integrated circuits that were embargoed to the Soviet Union.

VLSI?

Yes, the chess programs written by earthlings in the past have always been unable to win against humans. In addition to the backwardness of the algorithm, another big reason is that the computing hardware is too poor. The computing power can’t keep up. Later, under the care of Moore's Law, hardware became more and more powerful, and its ability to perform violent exhaustion became stronger and stronger. Finally, in 1997, IBM's "Deep Blue" defeated the world chess champion Kasparov, who was also a Russian. .

It’s not easy for people on Earth!

Yes, after the victory of Deep Blue, everyone thought that computer chess playing was over, and no one wanted to be thankless. IBM also disbanded the "Deep Blue" team. But in the mysterious In the East, there is a game called Go, which is much more complex than chess. Its search space is more than 10^170, which is more than the number of particles in the universe. It is impossible to rely only on evaluation functions and pruning.

Is this a piece of cake for our computers here? How did the people on earth solve it?

The people on earth proposed a method called Monte Carlo They often use the example of calculating the surface of a circle to introduce this method: draw a circle on the edge of a square, and then randomly throw sand grains into the square. When enough grains are thrown, start counting how many grains of sand fall. In a circle, divide this number by the total number of sand grains thrown and multiply it by the area of ​​the square, which is the area of ​​the circle.

​

Well, there are really some geniuses among the people on earth!

When they let the machine play Go, they also introduced the Monte Carlo algorithm to make moves where the probability is highest. One A company called Google developed a program called AlphaGo, which used Monte Carlo algorithms and deep learning, combined with powerful computing power, to actually defeat the top human chess player Lee Sedol!

So, humans have conquered all chess games?

No, they haven’t actually conquered it. I watched it quietly At first glance, AlphaGo is really a black box. People on earth can’t explain what is inside at all. Even using our planet’s technology to analyze it is difficult.

In fact, AlphaGo does not Emotion, without intelligence, has the pragmatic philosophy of victory and winning the game as its only goal, rather than the rationalism of pursuing winning strategies or optimal theories.

What people on earth really conquered was checkers. The Chinook checkers program defeated the then checkers champion in 1994. It was not until 2007 that people on earth theoretically proved that "as long as both sides of the game make no mistakes, it will eventually be a draw."

Well, it seems that this black box is not of much use!

No, no, people on earth are very enthusiastic now. They are trying to apply black boxes to all areas of human life, and have achieved good results in some areas, such as face recognition. They call this thing artificial intelligence now!

How could the artificial intelligence they developed be so weak? Send me the code to see!

9527 searched for a piece of code and sent it to the director.

Are all people on earth so bad at writing code?

No, no, I found this randomly. They also have good code, such as Linux.

The director scanned the code for a millisecond and immediately discovered the problem: Huh? Are there only three basic structures in a program: sequence, loop, and branch?

Yes, that is the case under the von Neumann architecture!

It seems that the earth’s technology tree has really grown crooked!

Should we intervene? Tell the people on earth the principles of computers we invented a hundred thousand years ago?

No, we can only observe. Interfering with the process of civilization violates the Cosmic Civilization Convention. You can continue to monitor it. ​

The above is the detailed content of Unfortunately, the development of artificial intelligence technology for people on earth has taken a wrong path. For more information, please follow other related articles on the PHP Chinese website!