An AI Learned Magic: the Gathering, Now Creates Thousands of New Cards 104
merbs writes: Reed Milewicz, a computer science researcher, wowed a major online Magic: The Gathering forum when he posted the results of an experiment to "teach" a weak AI to auto-generate Magic cards. Milewicz had trained a deep, recurrent neural network—a kind of statistical machine learning model designed to emulate the neural networks of animal brains—to "learn" the text of every Magic card currently in existence. Then he had it generate thousands of its own.
He shared a number of the bizarre "cards" his program had come up with, replete with their properly fantastical names ("Shring the Artist," "Mided Hied Parira's Scepter") and freshly invented abilities ("fuseback"). Players devoured—and cheered—the results.
He shared a number of the bizarre "cards" his program had come up with, replete with their properly fantastical names ("Shring the Artist," "Mided Hied Parira's Scepter") and freshly invented abilities ("fuseback"). Players devoured—and cheered—the results.
A bit disappointed (Score:2)
This seemed cooler than it actually is, as practically everything the program generates is completely nonsensical. As such, the end result does not seem special compared to everything else "AI's" have supposedly created in the past.
Hopefully the randomness hits home a couple of times and gives someone actually useful ideas.
Re: (Score:2, Insightful)
You severely underestimate just how much training goes in to a 2 year old's brain in order to identify trees. Not to mention the billions of years of the evolutionary equivalent to systems engineering that went in to the wetworks between his/her ears.
This guy spent a day dickering around with some existing programs and let it run overnight on ordinary computer and ended up with some interesting and quite non-trivial results.
Computers in any capacity have been around less than a century and they've come pret
Re: (Score:2)
Even when they weren't interesting, they were often quite entertaining. For example:
Hahaha... pre-school voice: "Very good, Billy! You get a cookie!" ;)
Re: (Score:1)
Agreed. That's why I call these systems a.i. (artificial ignorance) as opposed to A.I. (Actual Intelligence)
Re: (Score:1)
Agreed. That's why I call these systems a.i. (artificial ignorance) as opposed to A.I. (Actual Intelligence)
Or, to quote a few Science Fiction novels: A.S. = Artificial Stupid.
Although, that does seem a bit harsh. The poor machine is "dancing as fast as it can".
Disappointed Dipshits (Score:5, Interesting)
No one cares that you are not impressed, and no one cares about impressing you. Please don't bother posting this drivel.
Obviously this is not "real" intelligence. If and when that is developed, you can bet that it won't have anything to do with Magic cards. That you even expected that when reading this story means that you not only have no idea what AI research is all about, or much of an idea about programming in general, but also it speaks volumes about what your actual intelligence level is. Your maturity may also be called into question due to the content of your post.
You'd think with the level of tech expertise on here, you would have fewer people confusing the programming concept of AI with the science fiction concept of AI. If hard AI existed that's what the headline would be about. So far it does not; please refrain from polluting this forum with observations to that effect.
Re: (Score:1)
You don't need to act so high-and-mighty -- he's got a decent point. When I saw the headline I actually got excited. AI could be used to run thousands of deck-on-deck simulations to learn reams of information on different useful strategies. Instead he used it as a random name generator that pointlessly assembles cards. Now, combine the two concepts and you might actually be approaching the AI systems that are actually used to make the real cards in the first place.
Re: (Score:1)
He used a neural net, not a random name generator. Otherwise it would not be described as AI. Running deck-on-deck simulations is also not AI. Why is it that people that are so fundamentally clueless are compelled to post on this topic, especially after determining that it is not what they think it is?
This article is about an unusual application of a programming technique, mostly novel for its entertainment value. It's almost certainly not the best way to find new card mechanics. It has absolutely nothing t
Re: (Score:2)
Just a thought.
Re: (Score:1)
The post you responded to is a level-headed opinion of an average slashdotter. It wasn't rude. It even expresses some optimism.
You are rude. You insult other people. It is ironic how criticize someone else's maturity while maintaining absolutely none yourself. There is no redeeming value in your post. You seem like an angry child who can type.
Re: (Score:2)
"Oh, no it's not real, hard AI, it's something totally different which we just like calling AI even though there is no intelligence involved."
Just call it something else.
Re: (Score:2)
You need to come up with a more accurate name for "the programming concept of AI " that doesn't involve the term AI, if there is not any actual Intelligence involved.
Neural Networks (Score:2)
The term for this technique is 'neural network'. It's one of the oldest concepts in AI research. I don't know what planet you live on that "actual intelligence" is remotely close to being a reality. Doubly so because no one can even define what that is, and the results of all AI research to date are merely defining what intelligence is not. We do not need to invent a new term for people who are so ignorant of the field that they cannot tell it apart from science fiction. Even if "strong AI" existed, neural
Re: (Score:1)
I'm working on a system that can do things like that, but its Strong AI not weak, totally different. It dynamically adjusts and learns and has (will have) 'hyper-complex' 3D vision. The whole design ethos between strong and weak is totally different. - Strong essentially follows and reverse engineers the human / animal mind and brain. Strong is built around a 'dynamic core' or machine consciousness. Strong generally requires a high performance, custom designed and built multi-processor computer.
Lets face it
Re:A bit disappointed (Score:5, Funny)
I'm disappointed for entirely different reasons. I read, "An AI Learned Magic..." and thought, "Wow! What could that mean? Did it learn how illusionists perform their tricks? Are they claiming it somehow learned real magic? This should be interesting!"
And then I continued reading.
Re: (Score:2)
Sounds human-like to me.
Re:A bit disappointed (Score:5, Informative)
No. It made nonsensical cards early into its learning process.
Later on it made cards like this:
Re: (Score:3)
Well it's a good start.
But...
There are no Angels in green. They are all white, black and maybe red.
Cards like this only work as part of the whole if the rest of the set takes cards like this into account. Otherwise it's all random and unbalanced.
Re: (Score:2)
Well it's a good start.
But...
There are no Angels in green. They are all white, black and maybe red.
Are you sure about that? [wizards.com]
Re: (Score:2)
Well, he phrased it poorly. There are multicolored Angels that include Green (they're all also white, which has always been the preferred Angel color). There are no mono-colored Green Angels; there are three in black (and last seen with the reprint of Fallen Angel in Eighth Edition), and one in red (Akroma, Angel of Wrath) and one in blue (Illusory Angel). White, meanwhile, has *84* mono-colored angels.
Re: (Score:2)
Looks to me then we are due for a blue and green angel. Maybe the AI is not that dumb after all.
Re: (Score:2)
I can also add that *all* multicolor angels count white as one of their colors.
Re: (Score:2)
This seemed cooler than it actually is, as practically everything the program generates is completely nonsensical. As such, the end result does not seem special compared to everything else "AI's" have supposedly created in the past. Hopefully the randomness hits home a couple of times and gives someone actually useful ideas.
I watched a demo (can't find it now) of TempleOS [slashdot.org] where... God spoke to you!
Re: (Score:2)
This must be the first time the phrase "seemed cool" has been used in reference to Magic the Gathering. :)
Re: (Score:2)
Re: (Score:2)
Just think of the possibilities of Bayesian MTG cards... now instead of being full of GET TAXI CAB IN INDIA, the /. submission queue could be turned into an entertaining and enjoyable* game!
* for certain values of entertaining and enjoyable that may not apply to people who don't like CCGs
Re: (Score:2)
I was actually pretty impressed at how close it came to creating sensible cards, and pretty funny when it made small errors that made the cards absurd (such as the card with, "At the beginning of each player's upkeep, sacrifice a white Zombie creature").
Getting a computer to generate understandable language is an extremely difficult problem, and all neural networks have an issue with long-tail errors (that is, a small fraction of the results are always ridiculously inaccurate, no matter how good your neural
Re: (Score:2)
This seemed cooler than it actually is, as practically everything the program generates is completely nonsensical.
Partly because of this, I doubt that there is really much "AI" going on here. To me, this looks very much like simple probabilistic Markov chaining, with maybe a couple of rules to demark meaningful game elements that can be treated as a unit.
A bit more than that, maybe, but not much.
Re: (Score:1)
Sigh. (Score:4, Funny)
This is exactly why we need a moral framework for AI development.
Re: (Score:3)
Re: (Score:2)
that's not stoping Wizards from printing crap like Scion of Ugin and Comet Storm...
*grumbles*
Somewhere out there, a tree is working tirelessly to produce air for us to breathe, then it gets cut down and turned into crap bulk cards.
Re: (Score:2)
Uh, both of those cards are fine for casual. The reason Magic is awesome is cards the Spikes out there consider "jank" can be a lot of fun for casual magic players.
Re: (Score:2)
I think casual is where it gets less fun because if one casual player in a pool of several casual players pulls better cards than everyone else, especially at common and uncommon, pulling a Comet Storm in your rare slot in a pack of Modern Masters 2015 is that much worse.
I don't think Wizards is just going to sit by and let this happen though. My huge hope is that with standard going to two, two set blocks coming soon with Origins, this is going to make WotC R&D tighten up sets and have less chaff.
I thi
Re: (Score:1)
Oh shut up.
Re: (Score:2)
Re: (Score:2, Insightful)
-- C.S. Lewis
That is to say, yes. The opinion is changing as the kids who played pokemon and MTG grow up and continue to play, and the people who have apoplectic fits that these adults weren't swilling beer sitting in front of the tv 4 hours a day like "real adults" die off from heart attacks and strokes.
Re: (Score:2)
Draft nights are a pretty useful way to go out and be sociable even if you don't have a group of people to go with. So appeal++ for people who don't go out with friends most nights, I guess, but I wouldn't call ccg players social rejects.
Re: (Score:2)
As a programmer I don't have free time for games so I'm in the dark about them.
I don't see what the one has to do with the other. "As a programmer", I have friends who have been interested in various card games and RPGs for years, and I've also partaken every now and then.
Well I've always heard that these kind of games are classified as being for "social rejects" only.
Classified by whom? None of these games work well for someone who isn't social, since they can't be played alone.
Re: (Score:2)
Are you trying to be funny? Or are you just dumb?
The same question might be asked of you.
Re: (Score:2)
Why can't I ever think of these snappy comebacks until hours later.
Probably because you're anonymous.
Re: (Score:1)
Umm... They do on mine? I get notices in the upper-right telling me about messages, I read those messages, and in those messages I get comments from the ACs. Otherwise I would never know when they replied and I would be unlikely to reply to them - and I do.
Re: (Score:2)
A software system that's coded to handle many different cases is 100% artificial and 0% intelligence. But it still qualifies as artificial intelligence, since it does combine both aspects of AI.
All that means is that the term "AI" is totally meaningless. A digital alarm clock is 100% artificial and 0% intelligence too.
Re: (Score:2)
Every time you read about "AI" like this and you dig into the details, you find that the programmer set up so many parameters on it that it is nothing like AI.
I'm betting this one is no different.
Incorrect AC. OI (Organic Intelligence) has boundaries and parameters, so will AI. You can't escape it. Physics means there are limitations to all things in the physical world. Even an AI controlled one. With out rules, there would be no way to accomplish anything.
As for the posting that is referenced, I understand these games of chance which are similar to any other (poker, flipping coins, slot machines, etc all are based on a probability/chance framework). If an AI system designs the probability cards,
Re: (Score:2)
Get the AI's addicted to cardboard crack like MtG, so they will be distracted from killing us. Roko's Basilisk solved!
Re: (Score:2)
Re: (Score:2)
1
Artifact Creature - Dragon
For every round prior to summoning Roko's Basilisk, if you did not waste at least 1 mana during each prior round, you lose 1HP per round since the beginning of the game. If no player remembers how many rounds have been played, you lose.
99/99
Re: (Score:2)
To bad I didn't stop to think about what I was doing when I threw out my old magic cards I could of dusted off some of those alpha beta and filled up my check book.
Re: (Score:2)
I miss my dual lands. Once upon a time they were cheap...
Re: (Score:2)
Is it time to sell?
Re: (Score:2)
I'm far enough out of the loop that I have no idea about very recent market trends, just that they're higher now than ten years ago and seem relatively stable.
Re: (Score:2)
They are on that list of things I wish I hadn't thrown out along with an unopened millennium falcon model I tossed when I went off to college. The alpha beta and unlimited editions appear to be fetching some good prices if you happen to have a hoard of them you could make out like a bandit.
Re: (Score:2)
I hate to break it to you, but you are unlikely to get a Black Lotus in your random box. In almost all cases when people sell boxes of "random" cards they have taken out all the good ones to sell individually. Just sayin.
Re: (Score:2)
Mox Ruby | Mox Ruby | Mox Ruby | Mox Ruby | Mox Ruby
You just played five cards of the same type from your deck. You lose by default; you are only allowed a maximum of four cards of any name that is not a basic land.
waitbutwhy article (Score:2)
So with all the recent fuss over AI and some respectable folks being scared to death of it, I happened to stumble on this great article on waitbutwhy:
http://waitbutwhy.com/2015/01/... [waitbutwhy.com]
It's a long two parter, but well worth the read. If you want the tl;dr part, skip to part 2 and search for "Robotica". With that in mind, we're going to end up with a planet of mile-high stacks of Magic: The Gathering cards.
Re: (Score:1)
(there may be some modern-day humans still around, in something like a historical wildlife preserve.....those will be the children of people who can't let go of their familiar reference points in order to reach for something better. In order to fly, the bird must first release the branch.....you cowards who white-knuckle it can live in a zoo and fling poo at tourists).
What makes you think that we are not doing that now?
Re: (Score:2)
We are all going to be cyborgs, and it will be awesome!!!.
No diseases, no handicaps, cheap replacements for whatever gets damaged, superior instrumentality, limitless cognitive abilities, true space travel, mind-machine virtual reality....
It will be amazing. We will transcend every human limitation and become something new entirely.
*THAT* is how A.I. will destroy the human race. Not by genocide, but by helping us evolve into something entirely post human.
(there may be some modern-day humans still around, in something like a historical wildlife preserve.....those will be the children of people who can't let go of their familiar reference points in order to reach for something better. In order to fly, the bird must first release the branch.....you cowards who white-knuckle it can live in a zoo and fling poo at tourists).
Indeed, resistance is futile.
Disappointing (Score:5, Informative)
What I thought from the description is that a neural network was taught how to play Magic and somehow generated new cards by trying to play with them. Think for instance of a program that tries to come up with a new chess piece by coming up with a movement pattern, playing games with that piece and trying to figure out whether it'd be useful or interesting to play with.
This on the other hand looks like something like a markov chain generator. Amusing nonsense that can give humans fun ideas.
Re: (Score:3)
Re:Now Creates Thousands of New Cards (Score:5, Interesting)
Eh, play drafts in whatever the current standard set is. You'll run into a couple hundred cards maximum, all using the same few mechanics, and because everyone's trying to draft out of a very limited pool, you don't need universal knowledge...you just need to understand the game enough to build something with the pieces that are dropped in front of you. Plus it's cheaper and no asshole trick decks. Do it at the start of a block and there's a VERY small set of stuff to learn.
(sure it helps to be aware of what nasty tricks might be available, but it's really not that essential, and you can pick it up real fast)
As far as I'm concerned draft is the only fun way to play, haven't in a while now but still. Like ten bucks will get you a night of 5-10 games on the same level playing field as everyone else. Wizards' business model around draft games is to compete versus movies for friday night entertainment, and it's not really all that exploitative by comparison.
OK (Score:2)
What would be amusing would be to get to the point where a person could generate their own set, print them out with the approximate levels of rarity necessay for a set, and then have a draft with the random cards.
Who wouldn't like a card with MointainSpoink and Tromple?
Scenario Generator (Score:2)
It would be nice to see the same level of logic applied to scenario generators for games with large, persistent worlds. Skyrim and GTA games are what come to mind soonest for me. It would really pump up replay value. Probably neither of those games really needs it, but it would be a nice feature for a competitor.
AI Opportunities in MTG (Score:1)
There are a few interesting ways AI can be applied to the game of Magic the Gathering - this is one, and it's cool to see this guy's project, and the results.
I've done a little bit of thinking in this area, but more in the area of making an AI play with the goal of searching for new competitive deck lists.
I posted on my blog about my approach a while back: Getting Your Computer to Make an Awesome New Magic the Gathering Deck [hackshop.com]
Another interesting tack would be to see if you can write an agent to create single
Reminds me of my old D&D Random Character Crea (Score:1)
Next (Score:3)
His next research project is to make an AI that defends against copyright infringement lawsuits.
M:tG metadata (Score:2)
Huh. I wonder if you could do anything interesting by linking the card database to the various databases of card sellers -- price, stock and sales rate, number of editions in which the card appears.
knack for generating profoundly useless cards (Score:2)
I noticed that the network, now more fully trained, could generate meaningful, novel cards. However, it also had a knack for generating profoundly useless cards. Here are a few snippets from the output:
* When $THIS enters the battlefield, each creature you control loses trample until end of turn.
Not a bonus, but plenty of creatures have slightly negative effects if they cost less to summon than their positive traits might suggest.
* Whenever another creature enters the battlefield, you may tap two untapped Mountains you control.
Weird, but if you're prevented from tapping mana sources for some reason...
* 3, : Add 2 to your mana pool.
Useful if you're tricked into a large mana-burn situation. It effectively reduces all mana-burn down to 1.
* Legendary creatures can't attack unless its controller pays 2 for each Zombie you control.
Oddly specific, but not useless.
Re: (Score:2)
I noticed that the network, now more fully trained, could generate meaningful, novel cards. However, it also had a knack for generating profoundly useless cards. Here are a few snippets from the output: * When $THIS enters the battlefield, each creature you control loses trample until end of turn.
Not a bonus, but plenty of creatures have slightly negative effects if they cost less to summon than their positive traits might suggest.
True, but having your own creatures lose trample is unusual as a drawback to a card, it's a very very situational penalty.
* Whenever another creature enters the battlefield, you may tap two untapped Mountains you control.
Weird, but if you're prevented from tapping mana sources for some reason...
Due to a combination of how the stack works, and the majority of mana sources having the timing priority it's pretty much impossible to prevent a player from tapping a mana source for mana. The best you can really do is force the timing of it so that it's not useful for your opponent to tap a source for mana, for example if you use a spell or effect that would cause an opponent's land t