Designing Magic Cards using a Neural Network

Today I came across a fascinating article that actually made national news about a PhD candidate researching deep neural networks in computer science. This candidate designed to create such a network, input the data of all current Magic cards, and have this network design new cards based on this process.

According to the article, about 2 hours into the process, the network was already generating such fantastic cards as:

Slidshocking Krow - U
Creature - Dragon
Tromple,Mointainspalk
4/2
#Slidshocking Krow is ridiculously overpowered. A 4/2 for 1? In blue? With Mointainspalk AND Tromple? I see power creep is alive and well.

and

Grirerer Knishing - 4G
Instant - Arcane
Exile target creature you control.
#The price is a little steep on this one, but maybe it's worth it for the synergy with other Arcane spells...

Interestingly enough, by letting it run overnight, the network soon began to create actual meaningful and unique cards, such as:

Gravimite - 1(G/W)(G/W)
Creature - Dryad 1(G/W): Regenerate $THIS.
When Gravimite enters the battlefield, draw a card.
2/3
#I think this is a reinterpretation of Carven Caryatid.

or

Shring the Artist - 2BB
Legendary Creature - Cat
Flying
Whenever you cast a spell, you may return target creature card from your graveyard to your hand.
2/2

This of course amused me to no end. I think I actually giggled as I read some of the cards. But it gets better. Apparently a member of MtgSalvation named pickfifteen later created a Twitter account (named "RoboRosewater" naturally) that generates full cards using this network. It's definitely worth checking out.

Of course, the system still isn't perfect. It apparently also likes to generates cards like this:

I don't think R&D needs to worry about being replaced by this network in the near future. However, they may want to offer this guy a job...