5 Games Where AI Is Already Stronger Than Humans

Currently, the capabilities of Artificial Intelligence (AI) are worrying more and more observers. However, that has not always been the case. In games, she first had to prove herself by defeating world champions in their respective fields.

At the time, the main games in which the machines challenged the human being are the strategy games. The game of checkers and the game of go are examples. During the various clashes, hardly anyone, sometimes even the designer, could imagine the victory of artificial intelligence (IA). Yet the representatives of the machines were able to surprise the whole world. Today, the capabilities of these algorithms in games surpass often from afar those of an average human.

1 – Chess: the defeat of the great Kasparov

Years 1996 and 1997 were the scene of very important events in the world of chess. It is Garry Kasparov’s clash against Deep Bluewhich is the computer designed by IBM. As for Garry Kasparov, he is considered the greatest champion in chess history. The first year, Kasparov wins with a score of 4-2. Although the representative of Man won, the computer program still won a victory.

Garry Kasparov against an AI

“I could feel – I could feel – a new kind of intelligence from across the table. »

Garry Kasparov

The following year, Deeper Blue, the upgraded version of Deep Bluetook his revenge. IBM’s machine won with a score of 3.5 to 2.5. During this new confrontation, the computer won in two games and made three draws. This episode marks the first defeat of a world champion in a game against a computer system.

“People were following this game all over the world, people who had no interest in chess, people who only had an interest in this tale of man versus machine. »

ESPN reporter Jeremy Schaap

This last confrontation between the robot and Gary Kasparov brought together several million people. Indeed, it is an extremely significant event, especially in the recognition of the potential of artificial intelligence.

2 – Go: the superiority of AlphaGo AI

The game of go has existed for more than two millennia, it is one of the oldest games. In terms of complexity, it is more sophisticated compared to chess. However, the latter are a reference for evaluating the intelligence quotient (IQ) of Man. Therefore, the best go players are considered as extremely intelligent.

AI in the game of go

In chess, the Deep Blue artificial intelligence (AI), for example, processed a large number of moves to determine the best one to make. In this chinese board gameto do the same is simply impossible. Despite the simplicity of the rules, go offers a possibility of the order of about 10600. In comparison, the number of atoms in the universe is estimated at 1080it is simply insignificant.

However, this complexity of go has not prevented DeepMind from providing the capabilities needed to Alpha Go to dominate the human being. In 2016, he stunned several hundred million spectators in a match with a million dollars at stake. For good reason, artificial intelligence defeated Lee Sedolwhich earned him the honorary title of ninth dan. It is the highest existing grade in this discipline and artificial intelligence has earned it.

SEE ALSO: Warning: AI is now able to read our minds

Indeed, Lee Sedol is a professional player who is also ranked ninth dan. Despite the absence of a world championship in this discipline, he won 18 international titles between 2003 and 2015. These successes have earned him the status of the best human world player in the mid-2000s.

During the game between Lee Sedol and AlphaGothe latter showed two surprising qualities. Her creativity and his reading and assessment skills are of an impressive level. Due to this defeat, the player announced in 2019 that he was retiring.

“Even if I become number one, there is an entity that cannot be defeated. »

Lee Sedol

Furthermore, this artificial intelligence also have simultaneously faced five of the best players in the worldincluding Lee Sedol. The match ended in defeat for the humans. However, far from being devastated, most gamers enjoyed playing against AlphaGo.

3 – Poker: the AI ​​Libratus, an Ace of Poker

In poker, bluff and spontaneity are necessary. Needless to say, an artificial intelligence did not have such capabilities. In addition, unlike the previous two games, the data available to her are incomplete. Indeed, the system receives only incomplete informationbecause he does not know his opponent’s cards. Program artificial intelligence (IA) capable of deceive the adversary and act reasonably with little data was a big challenge.

However, a robot managed to upset the world of poker in 2017. The machine in question is called Balance and she is the product of the work of Tuomas Sandholm and Noam Brown. At that time, the former was a professor of computer science at Carnegie Mellon University, while the latter was his doctoral student. Besides, Libratus has received special training for no limit texas hold’em poker heads up. This last term means that the games are played face to face.

It’s in January 2017 what happened to the competition “Brain vs Artificial Intelligence”. She staged four of the best human players against Libratus. Dong Kim, Jason Les, Jimmy Chou and Daniel McAulay are the ones who were to face the latter. All four are well known players in the poker world. Among them, Dong Kim and Jason Les have already faced the previous version of this artificial intelligence, Claudico.

An AI playing poker

This last match took place in April 2015 and resulted in the defeat of the computer system. However, his successor, Libratus, managed to seize victory and it was not a simple victory. In effect, the big four players were totally helpless to the capabilities of this artificial intelligence.

“When I see the robot bluffing humans, I’m like, ‘I didn’t tell him to do that. I had no idea he was capable of it.” It’s satisfying to know that I’ve created something that can do that. »

Noam Brown

4 – Backgammon: Luigi Villa beats AI BKG 9.8

The backgammon is one of the games played on an apron using dice. Whether chance has its place in this game, strategy is much more important here. Indeed, the player has a large possibility of movements each time he rolls the dice. However, he cannot afford random actions, because he must also foresee those of his opponent.

AI in Backgammon

in 1979Luigi Villa has won the world championship of backgammon which took place in Monte Carlo. The day following this success, he faced BKG 9.8 computer system that Hans Berliner designed. Faced with this artificial intelligence, the world backgammon champion had no choice but to bow.

Berliner noted that Luigi Villa had less luck than BKG 9.8. According to him, the Villa’s level of play was higher. But unfortunately the artificial intelligence dice rolls were more advantageous. Thus, the match ended with a score of seven against one in favor of BKG 9.8.

5 – Checkers: Marion Tinsley remains undefeated

The checkers or english checkers is a variation of the international checkers game. Marion Tinley is the reigning world champion in this field. In fact, he is even considered the greatest checker player which has never seen the light of day. During his professional career, Tinsley only lost seven games. Among the seven, artificial intelligence Chinook deserves credit for being behind two of those losses.

Marion Tinsley has faced Chinook twicein 1992 and 1994. During the first meeting, the artificial intelligence and the representative of the human being participated in 39 games. The match ended with the Tinsley victory with a final score of 4-2.

Checkers, a game where the AI ​​is excellent

Unfortunately, no one will know if the artificial intelligence could win against its adversary. Indeed, during the revenge of the machine, Tinsley had to withdraw for health reasons. He was diagnosed with a pancreatic cancer causing his death seven months later. The meeting between Tinsley and Chinook ended with six draws and the victory of artificial intelligence. However, she triumphed ingloriously.

In 2007, Jonathan Schaeffer, one of the creators of Chinook, completely solved the checkers. He demonstrated that a perfectly executed game must end in a draw. At the same time, Schaeffer’s team made artificial intelligence capable of playing English checkers perfectly.

Similar Posts

Leave a Reply

Your email address will not be published. Required fields are marked *