Having the AI be able to play at all is easier only because nearly all cards' effects are in some way good whereas in chess, it is much easier to make a move which is actually BAD. However I suspect trying to find an optimal strategy in elements would be significantly more difficult since as I said, there are far more possible moves to choose from on each turn as well as not having complete information about the position(ie not knowing your opponent's hand or even what cards you will be drawing in the next few turns). While the AI may be able to play at a level which is at least halfway decent in most cases, how often do you see it do something completely stupid like playing a dune scorpion, immediately followed by quintessence?
Also, it would be a bit unfair to compare the difficulty of writing a chess program which can compete at a grandmaster level vs. an Elements AI playing at the level of an average player. And even then, the only AI level which presents a significant level of difficulty for a reasonably skilled player with a good deck is FG's and then ONLY because they are given significant advantages like double hp and draws, triple mark, larger deck etc. Making a chess engine that could be the world champion would also be fairly easy if it could make 2 moves at a time and move pawns up to 3 spaces forward