Will AI beat top Magic the Gathering human player before the end of 2028?
37
2.5k
2029
31%
chance

Resolves to Yes if an AI has the strength to beat world champions in Magic the Gathering before the end of 2028

If AI were able to beat the world champion in any one of the major format, Draft, Constructed or Sealed, it would count as a resolve Yes. The AI doesnt have to actually be participating in the world championship (it likely would not), but as long as the AI demonstrated consistent strength in beating current world champion in repeated games (70% win rate with 10+ games played) that would count as a Yes.

AI doesn't necessarily have to beat the current world champion. Beating any of the top 20 in the world within past 3 years is sufficient. The goal is to demonstrate that the AI has the strength to likely win against the world champion.

related market:

Get Ṁ600 play money
Sort by:
bought Ṁ50 NO

70% win rate against a world champion goes beyond just "superhuman" - it's not obvious to me that that's possible even with perfect play, there's just a lot of variance in Magic.

Heres my take on this one:

I believe if any team of researchers made a serious attempt to make a superhuman alpha-go style Magic player, they would easily succeed.

But the question is: Will any team make a serious attempt in the specified time frame? To me that seems unlikely. Magic has a TON of annoying rules and individual cards to program in even aside from any actual AI work and simply doesn't have enough prestige to be worth the time from anyone serious enough to succeed.

bought Ṁ60 YES

What does this mean exactly? Piloting the same deck? Does the AI have to build it's deck? Constructed or Sealed?

@MichaelWheatley
If AI were able to beat the world champion in any one of the major format, Draft, Constructed or Sealed, it would count as a resolve Yes. The AI doesnt have to actually be participating in the world championship (it likely would not), but as long as the AI demonstrated consistent strength in beating current world champion in repeated games, that would count as a Yes.

I don't think building the deck would be harder than strategically beating a top player in the game? Meta deck lists are mostly online anyway.

@AmmonLam What does "consistent strength" mean, given that Magic is a game with a fair amount of randomness? If the AI has a 60% win rate, is that enough? Does it have to be the current world champion specifically, or would another top-tier competitive player count? Also, can it be any Constructed format? What if there's a new format specifically designed to be more tractable to AI?

@sesquipedalianThaumaturge by "consistent strength" I mostly mean that winning one or two game isn't sufficient. winning 8 out of 10 games or 14 out of 20 games would be what I find enough to demonstrate "consistent strength"

It doesn't necessarily have to be the current world champion. I would say beating any of the top 20 in the world within past 3 years is sufficient. The goal is to demonstrate that the AI has the strength to likely win against the world champion.

A new format specifically designed to be more tractable to AI likely would not count, unless that format got so popular that it is becoming common among tier A tournaments.

@AmmonLam Is that the metric then? 70% win rate?

@NoRespect sure. 70% win rate with 10+ games played.

More related questions