Thing is, X-com was so far ahead of the curve in terms of what it offered that it didn't really need all that many big changes, tweaks and fixes in response to some of the problems coming up sure, and it's what Nucom kinda did but overall, the core of the game (the two modes feeding off of each other and making each other more interesting than they would be on their own) was and still is a damn fun and engaging expirience. And it's rather telling that when the first one was made, they didn't actually cut or majorly revamp any of the features because they kinda got it spot on in the first go.
As for this, it's a bit of a shift it seems, where greater focus is on the metamap part of the game and emergent gameplay that comes out of it while the TBT sees some benefits from it but it mostly sticks to the tried and tested formula. I don't see an issue with procedurally generated monsters, that's stupidly simple to implement (hell, Impossible Creatures could've had that implemented easily enough and that thing is almost 15 years old), what the real issue becomes, like others have pointed out, is how they do the adaptability part. What criteria will the AI use and how, will it track damage dealt or recieved, efficency of certain units or a certain combination of factors? If implemented poorly it could become something to game and cheese in a particular way, or it could become a brick wall after a certain point. The only way to fix it really is to combined procedural stuff with a bit of handcrafting it I guess, where you create certain archetypes and the game then does variations on them until it figures out which one works best trough a course of several missions.
Either way, this is gonna require a ton of playtesting before it becomes something really nice, hopefully there's a good period of public beta testing of those features to iron out all the possible kinks.