Ngl I wasn't going to buy any UNTIL some creators were talking about the cards. I didn't end up buying much, just a couple singles, but I might not have otherwise.
Meanwhile my Kykar, winds fury "Hey I just read how possibility storm and dovescape, two of my favorite cards, interact" deck is inconsistent, but fun. The mana costs of spells are often multipip so occasionally cards brick in my hand, but when I get so much value from each card I do cast, it works
Been goldfishing two jeskai decks, a colour combination I don't play often.
Gwen Stacy//Ghost Spider plays really smoothly. Lots of card advantage and good combat payoffs. I have a couple spider man cards but all of my "if I'm upgrading this is what I'm cutting" cards are spiders...
I guess what I'm trying to say is "The AI made to pass the SAT passed the SAT" isn't much of a headline. What is more interesting is what weights in learning actually helped achieve it.
Or more likely, imagine a deck of cards, you reveal the top card. If it's not a 7, you remove it, then shuffle and repeat. If it is a seven you just shuffle and repeat. It shouldn't be surprising that eventually all you start pulling is 7s
