In the era of artificial intelligence, can guessing boxing be fair?

Guessing boxing has been used for a long time in decision-making and there is not much thinking in life. It is generally believed that the probability of players winning should be the same, which is about 30%. In the long run, this makes it possible for any player to win, draw or lose. This is the so-called mixed strategy Nash equilibrium, in which each participant chooses three actions with equal probability in each round.

But this may not be the case. The research team from Zhejiang University found that the strategies of real players seem random, but in fact they are composed of predictable patterns, which cunning opponents can use to gain important advantages.

Specifically, the research team developed an artificial intelligence model based on N-order Markov chain-Multi-AI, which means that Multi-AI has memory and can be traced back to N.

Historical state and use it. Researchers combine individual models to cope with the different personalities and strategies of human players.

If human players win in succession, it will promote multi-AI.

Turn to other artificial intelligence models for better solutions. If the human player fails continuously, it is likely to change the strategy or break the previous punching rules, and then multiple AI can be adjusted accordingly.

This means different game strategies. In other words, the multi-AI model emphasizes the timely regulation of personality differences and punching strategies among different players, and selects the most suitable game strategy at present.

The final experimental results show that after 300 rounds of fighting with 52 human players, artificial intelligence defeated 95% of the players. Because the rules of the game are to win +2 points, draw+1 point, without losing points, and all participants know that the winner will get a monetary reward. The higher the total score, the more money you win, so it is extremely unlikely that players will deliberately release water or choose randomly.

Even so, artificial intelligence has defeated human beings. In the most disparity contest, artificial intelligence won 198 victories, 55 draws and lost only 47 games, with a winning rate four times higher than that of human opponents.

Of course, this latest research result of guessing boxing is not only a powerful artificial intelligence of guessing boxing, but also a powerful model analyst of circular checks and balances, which means that human competitive behavior does have laws to follow, and these laws can be utilized by using appropriate simple models. Researchers believe that the artificial intelligence model is expected to be extended to other game scenarios in the future, such as predicting competitors' next moves, planning more effective campaign strategies, or formulating more favorable pricing schemes.