I agree with the others who say that this is the most playable initial release so far.
I didn't play the original Civilization, but I've been around for the launch from Civ 2 on, and all of them had many problems in their vanilla release (including Civ 4!). Problems with the UI, basic game mechanics, and of course balance have affected all of the releases, especially for the top-level players. One thing to keep in mind is the tremendous explosion in high-level players that happened in Civ 4. That was finally a game that you could spend thousands of hours on, get deeply involved in things like balance questions, and have a community to share in. All those experts mean that problem discovery is happening much faster in Civ 5 and 6 than it did in previous releases. Hopefully that (along with mod--abilty) means that problem solutions will occur in a shorter time period.
I think there's no question that Civ 6 has the potential to pass Civ 4 for most folks, but not for all. They are different style games that appeal to different folks. I personally got bored with Civ 4 playing less hours than Civ 5 (including all updates of each). Civ 4 is a much simpler strategic game than what Civ 5 tried to be, but it was a much richer mid-level tactical game. There were only a couple of paths of victory at the top-levels - the joy and challenge come from good implementations of those paths. The lousy AI in Civ 4 doesn't matter as much because it is a simpler game, and easier to give AI buffs without affecting overall game play. The better, but still lousy, AI in Civ 5 affected game play much more because of the difficulties of both high level strategy (which improved with updates) and the low-level tactics.
AI in Civ 6 will be interesting. It appears as though there is much more freedom for AI buffs to handle the greatly increased complexity. Even at the tactical level, the individual unit strength dependency on difficulty level will be nice (corresponds well with increased units in stacks as the Civ 4 buff).
What will really be interesting is if there's enough hooks for users to write their own AI routines. I spent some months investigating freeciv (open source Civ 2 implementation) but gave up due to real-life interference. Even there, it wasn't going to be easy to modularise the AI enough. I haven't checked into what Civ 6 offers at all; just enjoying the play for now!