A few days ago, artificial intelligence (AI) once again defeated humans.
This time, instead of playing checkers, chess or Go, StarCraft, GT Racing or Olympiad, the AI played a more entertaining card game– bridge.
In 1942, a bridge club at Schumer College in the United States | Wikipedia
In Paris, France, an artificial intelligence called Nook defeated eight bridge world champions in a bridge competition held last week, according to the British newspaper The Guardian.
This victory is a new milestone for the AI industry, as Nook has to react to the actions of several other human players while using incomplete information, a situation that is closer to "making human-like decisions."
Although artificial intelligence has previously defeated humans in chess and Go, at that time, artificial intelligence players only had one opponent at a time, and both humans and machines had all the information.
Explainable artificial intelligence
Bridge, a card game that wins the pier with skill, belongs to the game of eating piers, is an extension of Whistler (another card game), one of the most popular card games in the world, especially popular among the elderly population.
Bridge is played by two pairs of four people at a square table, with partners sitting face to face at opposite ends of the table.
A section of bridge contains many decks, and the order in which a deck is played is to deal the cards first, then to decide the contract, then to play the cards, and finally to register the results of the cards. The goal of a deck of cards is to achieve as good a result as possible with the cards handed out in the hand.
This time, in a bridge match called "The NukkAI Challenge," Nook, who had the same poker cards and the same opponent as the human champion, outperformed the human champion on 67 of the 80 matches.
The contest screen | YouTube
In response, Véronique Ventos, an artificial intelligence researcher and one of the co-founders of NukkAI, said that NooK is a "new generation of artificial intelligence" because it can explain when making decisions. "In bridge, if you don't explain, you can't keep playing."
Stephen Muggleton, a professor at Imperial College London, said the victory "represents a fundamental and important advance in the field of artificial intelligence".
Image source| The Guardian
In recent years, interpretability has been a hot topic in the field of artificial intelligence.
Most of the advances we hear about machine learning today are based on black-box systems like AlphaGo, where AI can't explain to humans how decisions are made.
NooK, on the other hand, represents a "white box" or "neural symbol" approach that learns not by playing billions of rounds of games, but by first learning the rules of the game and then improving the game through practice, based on a combination of rules and deep learning systems that are closer to humans.
Michael Littman, a professor of computer science at Brown University, said that even if a person or ai-intelligence can't explain what they're doing in words, their behavior needs to be "clear and understandable," which is crucial for areas such as health and engineering, such as self-driving cars about to pass through intersections, which need to be able to read the behavior of other vehicles.
But Littman also said that the process of this battle of people also has shortcomings, because there is no bidding (also known as calling) process in the whole game process, and bidding is the most interesting part of bridge communication and deception.
But in the view of Nevena Senior, one of the bridge world champions and one of NooK's opponents this time, NooK is better at interpreting opponents than humans, and can take advantage of opponents' mistakes, and the creators of NooK have done a "great" job.
"It's something that humans do only after they've accumulated enough experience. I was pleasantly surprised that robots could mimic human skills. ”
"Man-machine game" big inventory
Since the birth of artificial intelligence, the PK between it and humans has never stopped.
Thirty years ago, in 1992, the international checkers artificial intelligence program Chinook first challenged Tinsley, known as the "best in the world" checkers world champion. Regrettably, however, Chinook lost.
But two years later, Chinook regained his strength and challenged Tinsley again, and after six draws between the two sides, Tinsley retired due to illness and Chinook won the championship.
This is the first time in the history of artificial intelligence that it has won an official world championship in a competitive game.
But Chinook's progress did not stop, and in April 2007, Chinook's ultimate program was introduced, becoming an undefeated presence in the field of checkers.
In the winter of 1996, the first chess man-machine battle was played out in Philadelphia, USA. IBM's supercomputer "Deep Blue" played 6 rounds of human-machine game with chess king Kasparov. In the end, Kasparov defeated the "Deep Blue" by an aggregate score of 4:2.
This time, humanity defended its dignity.
However, just a year later, in May 1997, the improved supercomputer Deep Blue once again challenged Kasparov. At the end of 6 rounds, the comeback "Deep Blue" defeated the chess king with a record of two wins, one loss and three draws.
This is the first time that ai-powered intelligence has beaten a chess world champion within a standard match time frame.
But today's view, "deep blue" is not intelligent enough, its flaw is no intuition, does not have a real "intelligent soul", can only rely on super computing power to make up for the defects in analytical thinking.
It wasn't until March 2016 that another "man-machine war" shocked the world.
AlphaGo, developed by Google's artificial intelligence company DeepMind, defeated the then Go world champion Lee Sedol by a score of 4:1, becoming the first artificial intelligence to defeat the Go world champion.
This event is considered a milestone in the true sense of the ai industry.
Technically, by using a method that combines Monte Carlo tree search with two deep neural networks, AlphaGo can learn spontaneously like a human brain, perform analytical training, and continuously learn to improve chess power.
A year later, at the Wuzhen Go Summit in China, AlphaGo defeated the world's no. 1 World Go champion Ke Jie with an aggregate score of 3:0.
Since then, in areas such as esports, programming, and life sciences, various types of artificial intelligence have also sprung up.
For example, on December 10, 2018, DeepMind's AlphaStar for the real-time strategy game StarCraft destroyed 99.8% of the world's professional players.
In February, DeepMind created an AI system called AlphaCode and declared that its programming capabilities could compete with those of the average human programmer.
In the same month, the artificial intelligence program developed by Sony's artificial intelligence team also defeated the world champion human player in the racing game "GT Racing".
Today's artificial intelligence has penetrated into all aspects of our daily life, learning and work, and has the ability to match or even surpass humans in some specific tasks.
Therefore, some people will worry that the increasingly capable artificial intelligence system will not evolve its own consciousness one day in the future, and then surpass humans and even do more terrible things.
In fact, the advantage of artificial intelligence lies in its super memory, computing power, etc., and its ability to work endlessly under extreme conditions. At this point, current AI does outperform humans in many areas.
But from a technical point of view, the current artificial intelligence system is self-learning or artificial correction under the given data and specified rules, and can solve specific problems under specific parameters. Once a variable is introduced, the system may collapse or a new model needs to be created, far from the scene depicted in a science fiction movie.
Both humans and artificial intelligence have different capabilities and advantages. Artificial intelligence cannot have many unique human abilities and characteristics, such as emotional awareness and creativity, and humans cannot have super memory and computing power and the ability to work endlessly like artificial intelligence.
So the real concern is how to make AI work better with humans.
In any case, it is undeniable that AI has indeed changed some things.
Reference Links
https://www.theguardian.com/technology/2022/mar/29/artificial-intelligence-beats-eight-world-champions-at-bridge
https://en.wikipedia.org/wiki/Contract_bridge
https://challenge.nukk.ai/
Author: Yang Xiao