[GH-ISSUE #622] Using DeepSeek AI to create an AI for a card game #178

New issue

Open

opened 2026-10-26 21:04:21 +00:00 by bsakaguchi · 1 comment

bsakaguchi commented

2026-10-26 21:04:21 +00:00

Owner

Originally created by @shadowfox87 on GitHub (Jan 31, 2025).
Original GitHub issue: https://github.com/deepseek-ai/DeepSeek-Coder/issues/622

AI is evolving rapidly. We have World Chess and Go champions that have been defeated by AI. Yet, trading card games like Magic the Gathering and Yu-Gi-Oh still have yet to make an AI that can do this. This is because of the complexity of a card game that has so many different types of cards that do different things and the randomness factor which makes each game different from its past. This is unlike Go and Chess where all pieces are always static at the start of the game and have the same function.

We have a lot of data of a lot of games from several thousand players using the best decks. This data is in the form of a json that has every move every turn, the action, the card that is used and what happened. We want to be able to see there is a way to train an AI to write an AI script using this data for a specific deck of cards and that deck's duels.

If interested, contact me. I don't have the GPU or computing power to do this myself.

Originally created by @shadowfox87 on GitHub (Jan 31, 2025). Original GitHub issue: https://github.com/deepseek-ai/DeepSeek-Coder/issues/622 AI is evolving rapidly. We have World Chess and Go champions that have been defeated by AI. Yet, trading card games like Magic the Gathering and Yu-Gi-Oh still have yet to make an AI that can do this. This is because of the complexity of a card game that has so many different types of cards that do different things and the randomness factor which makes each game different from its past. This is unlike Go and Chess where all pieces are always static at the start of the game and have the same function. We have a lot of data of a lot of games from several thousand players using the best decks. This data is in the form of a json that has every move every turn, the action, the card that is used and what happened. We want to be able to see there is a way to train an AI to write an AI script using this data for a specific deck of cards and that deck's duels. If interested, contact me. I don't have the GPU or computing power to do this myself.

bsakaguchi commented

2026-10-26 21:04:22 +00:00

Author

Owner

@alelazcano commented on GitHub (Apr 2, 2025):

@shadowfox87 see this: https://www.youtube.com/watch?v=GIJ7zr4sYx4 Google Alpha Go ;)

@alelazcano commented on GitHub (Apr 2, 2025): @shadowfox87 see this: https://www.youtube.com/watch?v=GIJ7zr4sYx4 Google Alpha Go ;)