Our AI writing assistant, WriteUp, can assist you in easily writing any text. Click here to experience its capabilities.

Now AI can outmaneuver you at both Stratego and Diplomacy

View Original View Raw

Summary

This article discusses the recent advances in artificial intelligence, specifically in the area of board games. AI has long ago surpassed human capability in chess and Go, but more complex games like Stratego and Diplomacy have presented a challenge. However, recent advances have resulted in AI that can outmaneuver humans at both Stratego and Diplomacy. The key difference between these games and chess/Go is that they are based on imperfect information, requiring skills like guessing what the opponent is thinking and making moves that accommodate and hopefully upset those assumptions.

Q&As

What are the two games AI recently became good at?
The two games AI recently became good at are Stratego and Diplomacy.

What is the crucial difference between these games and ones like chess or Go?
The crucial difference between these games and ones like chess or Go is that they are games of strategy based on imperfect information.

How does the AI play Stratego?
The AI plays Stratego by making moves that cannot be exploited or predicted.

How does the AI play Diplomacy?
The AI plays Diplomacy by making predictions about what other players will do and then forming an intent for itself and its partner.

What are the implications of these findings?
The implications of these findings are that AI is becoming good at making decisions based on incomplete information, which has implications for many real-world situations.

AI Comments

👍 This is amazing! I can't believe AI is becoming so good at strategy games!

👎 This is terrible! Soon AI will be able to outwit us at everything!

AI Discussion

Me: It's about how AI can now outmaneuver humans at Stratego and Diplomacy.

Friend: That's interesting. I didn't know that AI was that advanced.

Me: Yeah, it's pretty amazing. The article talks about how these games are different from chess and Go because they're based on imperfect information.

Friend: That makes sense. I can see how that would be a challenge for AI.

Me: Yeah, the article talks about how the AI has to be able to bluff and convince the other player of something.

Friend: That's definitely something that humans are good at.

Me: Yeah, the article talks about how the AI is still a work in progress, but it's pretty amazing that it can be competitive at all.

Action items

Read the paper on Regularised Nash Dynamics.
Try playing Stratego and Diplomacy against DeepNash and Cicero.
Use the techniques from DeepNash and Cicero to improve your own game play.

Technical terms

DeepMind: an artificial intelligence research laboratory founded by Google in London, England.
DeepNash: a computer algorithm designed to play the game of Stratego.
Regularised Nash Dynamics: a method of game theory that is used to analyze and predict the behavior of players in games of strategy.
Cicero: a computer algorithm designed to play the game of Diplomacy.