Nowadays, card games are a very widespread entertainment product on the market. They are played by different categories of people of different ages, both for entertainment and for professional, commercial purposes. Casino is available in every corner of the world. It can be a casino in Poland or in any other European and non-European country.
There, you will receive dozens of offers of online casino products, and you will be able to play against real people or against artificial intelligence (AI). It’s possible because scientists and IT developers have managed to teach neural networks to play casino games very well.
What is the role of a neural network?
Despite the uncertainty and significant influence of luck, the so-called “lucky deck,” mathematics, as well as accurate calculations, play a significant role in the gambling sphere.
Even a completely and utterly unlucky person can easily master any card game and have a chance of winning if they know mathematics and are able to calculate and predict their and others’ moves.
A neural network tries to reproduce the way the human brain functions. Everything is done in this direction with the help of mathematical models. AI can be taught with external help. This helps create an algorithm for casino game providers to create a sufficient playing background for all casino players, regardless of the platform they choose.
The creation of poker-bot
Recently, the research group of the popular social network Facebook presented a general bot for artificial intelligence (AI). The new algorithm was named ReBeL. It could easily play both chess and poker. The developers expect that the new algorithm will work well in a number of more complex games.
The ReBeL’s developers explain that the program will not act primitively. Instead, it will be able to negotiate, detect fraud, and be responsible for cyber security.
The AI-based algorithm is not an innovation, but the special feature of ReBeL is that the system is able to learn itself and later imitate the game of the world’s players. This has been confirmed by numerous tests. During tests against the algorithm, real world-class professionals played poker. Three players can be mentioned from the list:
- Chris “Jesus” Ferguson (2000 World Series of Poker champion, winner of five WSOP bracelets);
- Greg Merson (2012 World Series of Poker champion);
- Darren Elias (four-time World Poker Tour champion).
ReBel’s Experiment
Two formats were chosen for the ReBeL performance experiment: five people played with one AI at a table, and one person played with five copies of the AI at a table.
In each of these options, six players started the game with ten thousand chips before each round. In the case of a five-player game against the AI, the latter earned an average of $1,000 per hour, which is a very good result.
This result shows better outcomes as compared to a situation when professional gamblers play against amateurs and professional gamers at once. When playing five copies of the AI against one player, the numbers were also in favor of the players.
What are the Outcomes?
In general, the algorithm is still weaker than real players because the professionals can quickly find vulnerabilities in the ReBeL game and ultimately win. Ferguson concluded after the experiment that ReBeL was a very tough opponent to play against.
As a result, it was possible to establish that the AI is still not up to the level of professional players and sometimes makes predictable decisions.
But in general, the developers see its progress and hope to improve the algorithm. But what we see is that neural networks can be taught to play casino games and the results of its potential success are quite promising.
