Artificial intelligence beats eight world champions at bridge
theguardian.com
theguardian.com
Final score card:
https://youtu.be/DHpeHCsojbI?t=15886
> Bot: 6147, humans: 5238
One of the human players, Thomas Bessis, speaks about playing against the bot (before the end of the challenge, when the final results were not yet known; in French):
https://youtu.be/DHpeHCsojbI?t=12173
Another of the human players, Anna Gulevich speaking about playing against the bot after the end of the challenge (in English):
https://youtu.be/DHpeHCsojbI?t=16703
Closing speeches by NukkAI:
https://youtu.be/DHpeHCsojbI?t=18472
The bot, Nukk, was created using Inductive Logic Programming (ILP) and Probabilistic Inductive Logic Programming (PILP).
ILP2018 paper by Legras, Rouveirol and Ventos on developing an AI player for Bridge using ILP:
https://link.springer.com/chapter/10.1007/978-3-319-99960-9_...
Each champion played their own and their “dummy” partner’s cards against a pair of opponents. These opponents were the best robot champions in the world to date – robots that have won many robot competitions but that are universally acknowledged to be nowhere near as good as expert human players.
Seems like two pretty hefty caveats. Why not pit their best AI against humans directly? And can someone with a better understanding of the game explain how important the initial bidding round is?
I confess I don't understand Bridge (where's my lands? Do I attack the opponent, or their cards?). Here's an older report on the state of computer Bridge:
https://cs.nyu.edu/~pbethe/bridgeReview200908.pdf
Should answer some of your questions.