Continue against AI after opponent concedes
Allow players to continue against AI after a human opponent concedes
I would like to suggest a feature where, when a human opponent concedes a game, MTG Arena gives the remaining player the option to continue playing against an AI opponent from the exact game state where the opponent conceded.
For example:
Opponent concedes → Match win is immediately recorded → “Continue against AI?” → Yes → AI takes control of the opponent’s existing board and hand/game state.
The original human match would still be completely finished and the win would be recorded normally. The AI continuation would simply be an optional, separate sandbox continuation and would have no effect on Ranked rating, matchmaking, rewards, statistics, or the result of the original game.
This would be particularly useful when:
- A player has assembled a complicated combo and wants to actually play it out.
- An opponent concedes immediately before the player can demonstrate what their deck is capable of.
- Someone is testing a new deck and wants to see how a particular board state develops.
- A player wants to practice sequencing or understand how their deck performs from a particular position.
- A player has spent several minutes setting up a board and would rather not have the game simply end without being able to use it.
The AI would not necessarily need to be perfect. Even the existing Sparky AI would be useful for this purpose. A stronger AI in the future would make the feature even better.
Ideally, Arena could offer two choices when an opponent concedes:
“End Game” — finish normally.
“Continue Against AI” — preserve the current game state and let the AI take over the opponent’s side.
This would not undermine competitive integrity because the human match would already have ended and the result would already have been recorded. It would simply give the winning player an optional way to continue playing with the board state they had earned.
Players have already discussed wanting this functionality in the MTG Arena community, so I believe there is genuine interest in it.
I think this could be a relatively simple value-add feature that would make Arena considerably more enjoyable, particularly for players who enjoy combo decks, experimentation and deck testing.
-
Adzomatic#36376
commented
Thanks @gWhiz#20143 for the comment, I think there may be a misunderstanding of what I’m suggesting. The AI continuation would not be a continuation of the server-side game or an AI player making decisions on the live match after someone concedes.
The human match would end immediately when the opponent concedes, and the win would be recorded exactly as it is now. The AI version would then be a local, client-side instance of that specific game state, running only on the remaining player’s device. It would essentially be a sandbox copy of the game at the exact moment the opponent conceded.
So concerns about the AI revealing the opponent’s future draws, searching their deck, revealing their hand, etc. could be addressed without requiring any of that information to be sent to the other player. The local AI would only be operating on the game information needed to recreate that particular game state.
If the concern is specifically that the winning player could potentially see information about cards that were hidden when the opponent conceded, I think that could be handled through a settings preference.
For example, each player could have a setting such as:
“Allow opponents to continue my conceded games against AI” — ON/OFF
If I have it ON, an opponent who wants to continue can play against an AI simulation of my deck/game state after I concede.
If I have it OFF, the game simply ends normally for my opponent.
I’m also not suggesting this needs to run on Wizards’ servers or affect matchmaking, Ranked, rewards, statistics, or the original game result. It’s specifically intended as a local sandbox continuation of the particular game that just ended.
I agree that it shouldn’t create infinite loops or excessive resource usage. Those would need appropriate safeguards, but they’re implementation considerations rather than fundamental problems with the concept.
The purpose isn’t to replace playing another game. It’s for situations where you’ve spent 10–20 minutes setting up a complicated board/combo and your opponent concedes just before you get to execute it. Being able to optionally play out that exact position against an AI would be a useful additional tool, particularly for testing and learning.
-
gWhiz#20143
commented
Is this to say that AI bot would make a choice, or draw a card, or reveal a hand, or mill remainder of deck, or allow search through opponents deck after the opponent has chosen to quit rather that give up any further intel?
If any of the above or anything like it are the case, I would strongly disagree with this approach. We may concede because we don't want to show what we would do, or what the next card is or what ever else may be in our decks or hands from there.
Or even worse, get stuck in an infinite loop scenario or other long, graphic or resource intensive occurrence that may crash our computer or perhaps the server if many players are doing this at once.
Sounds that it would also cause even more code bloat and memory leaks than are already in MTGA to begin with. It seems that just playing more games to practice would outweigh any benefits of this idea.