OK, so granted I'm totally new to the forums and have only put about 10 hours into DW, but I really really feel like there's a big issue here that's going unaddressed. That is: while there are obviously flaws (big ones) in the AI, I don't think that overhauling the AI is necessarily the path of least resistance, nor the best way of fixing the problem (read: making for a better end-user experience).
Most of the line of thinking has been that we humans have found optimal ways of playing that the AI doesn't take into account. Therefore, we humans have a huge edge over the AI. Therefore, the AI should be patched to play in those optimal ways. But as noted, that's a huge undertaking, possibly beyond the scope of anything short of a DW2. However, even if we could patch the AI up to at least closer-to-human levels, I still don't think that would fix the bigger problem.
What's the bigger problem? From how I see it, the bigger problem is that these optimal strategies exist in the first place, or at least that they're as powerful as they are. At their core, games are a series of interesting, meaningful choices. If the choices are meaningful but not interesting, the game suffers for it. Most of the strategies players can use to give a huge edge over the AI (focusing on one weapon type only, private sector money exploitation, 0% tax, et al) seem to me to be meaningful, but thoroughly uninteresting. Where's the tradeoff, other than the patience to micro your designs and colonies? Granted, I haven't gotten to that level of play yet, but if these strategies are as powerful as it sounds like, they detract from that element of choice that's critical to a game.
So what if we looked at this not from the perspective of "How can we make the AI abuse the system as well as humans can?" and instead asked "How can we bring some balance to the vastly overpowered min-max strategies?" It seems that would be easier than an AI overhaul, and it could easily do just as much or more to ratchet up AI competence. And while I may be alone in this, I think dialing back the power of min-maxing/munchkining/what-have-you would be good for the player experience.
For example, one thing that's been thwarting Ice this whole time is how hard it is to get the AI to ramrod one type of weapons tech. But couldn't this also be addressed by balancing the strategy of focusing on one tech line only against having access to a broader range of weapons? I'm not saying to shoot for a GalCiv-esque rock-paper-scissors system, but you could conceivably come up with a system that balances the raw power of highly refined weapons tech against having 2 or 3 weapons that aren't quite as well developed.
Granted, since it looks like the AI currently goes with an "all of the above (poorly)" priority set, you'd still need to tweak the AI. And of course to really let the AI capitalize on that system, you'd need to program it to constantly evaluate what sort of weapons and countermeasures it's up against, which would be tough to program and taxing on higher speeds.
Same thing could be said about 0% tax. As it stands, the AI doesn't abuse it, but I think it would be better to balance out 0% tax than to make the AI abuse it as much as players can. As another poster noted, you could implement very realistic penalties for jumping tax more than a few % at a time. That would be a good place to start. I'd also say that the growth rate bonus could be brought down as well. Not so far that 0% is a poor strategy, but far enough that it's not the only strategy if you want to play well. Not even "optimally," just "well."
But like I said, I'm a total DW noob and don't know anything. So feel free to correct/ignore me as you see fit.