First up, excellent news on the AI stuff. And I'm glad that you're consulting with the community.
We'll save the wheel discussion for the end, as I suspect you're going to get a lot of resistance on that. When it comes to balance, the problem is simply that there's too much of everything. That's all. The bonuses from buildings are too high in pretty much all cases compared to the cost of, well, anything. Just reducing pretty much all values by 30-50% will help.
Consider the humble factory. Currently, factories progress as 25%,30%, 40%, 55%, 75%. These totals are simply too high, when a factory planet might have anything upto 15-20 factories, this gives vast, 1000%+ bonuses to output on factories by endgame. This is why you have massive overproduction. Just look at the bonuses you get from relics - a top-level archaeology lab gives +40% to a stat EMPIRE-WIDE PER RELIC. If I have 5 manufacturing relics, I get 200% bonus manufacturing on every world... This is the kind of manu bonus that a size 10 industry world should be aiming for in total, not something handed out globally.
Combined with this are farms; these are also too powerful when 1 pop = 1 prod. We see players here with 150+ populations, covered in factories and hitting insane totals. If populations are to be allowed to get this high, then nerf production from population.
If we have production totals of 150, then bonuses from factories need to be nerfed massively. If we have factory totals of 1000%+, then food bonuses needs to be nerfed. It's pretty simple, tbh; pick what you want the highest manu per turn from a single planet to be (say, 500). Then decide how big you want populations to get - say 50. Now we just aim to make top-tier factories only produce enough industry to get planets up to that level - in this case, 20 factories producing 1000% bonus. Ergo, each top-tier factory gives 50%. Reduce everything else in line - so the factories go 10,20,30,40,50% rather than the present 25,30,40,55,75. This also has the advantage that if the AI chooses to upgrade factories rather than building out across the planet, it's not gimping itself.
I prefer the option of nerfing factories/labs/markets for this, as it reduces the difference between low-manu planets and high-manu planets - they can be 4-5 times as powerful as other planets rather than 15-20 times as powerful. This allows you to reduce the difference in timing between research world upgrades and industry world upgrades. But you could equally put it the other way - nerf farms to only increase by 2 food /level, or reduce population's production output by 50%.
In sum, if we're talking balance, we need to be working with smaller numbers. 5% increase per level in output is plenty. Top-level factories which are even just twice as good as bottom-level ones are still going to be plentifully powerful when you can have 10 of them.
As to the AI. I know you're not a big fan of scripted AI. I'm not either, tbh, despite scripting it. I'd much rather you produced some amazing stuff in the C. HOWEVER, that said, I don't think you're going to produce something amazing in the C in next two months. I have faith that 2 or 3 years down the line, the AI will be awesome; I don't have much faith that 2 or 3 months down the line you will have brought it up to a standard where it can beat the human reliably without bonuses. You have other things to do at the same time, and people rely on you doing those things so they can get paid and feed their families. So what I suggest is that we both hold our noses and you go ahead and put in some more script triggers for the scripted parts of the AI for now, in the full intention of getting rid of them later, so that in the short term we can script up something a bit more effective. It's a placeholder, but it needs to be a place holder for a long time while different modules are moved from the scripts to the full AI. Shifting stuff like the planet build queues to AI, then the blueprints etc will give incremental increases, but we need the AI to be able to manage it's economy effectively in the mean time - and that means we need more script triggers to tell it when to do things. Hell, you could even hold a forum competition - free SD games for life for the first person to script an AI that beats Brad. That way, you not only get a bunch of cool scripts to add into the game, but you also galvanize the community and drive involvement. I'm willing to bet I know 5 or 6 modders who would start scripting tonight.
Now, onto the death of the wheel... I understand why you want to do this; Paul, in fact, just outlined it on the stream. And I don't think it's necessary. There's been many suggestions for ways to make micro-managing the wheel less problematic - planet grouping is a forum favourite - so eliminating micro isn't the main reasoning here. The main driver of this is that players can achieve massive output from using it. But I suggest this is a balance issue - not a systemic one. I think that a significant re-balancing of the economy in a generally downward direction would eliminate the game-breakingly vast outputs that some players are getting. Combine that with teaching the AI to specialize (which is already present in proto-form in the governors). Whether we like it or not, the wheel is an integral part of the economic game in GC3; unless you re-implement something entirely different (like abandoning population=production and returning to the GC2-style money economy in it's entirety, or adopting Petri's population = % bonus and factories give flat bonuses) the wheel needs to stay.
I'd suggest that, before you go to the effort of removing the wheel and thinking up a replacement, you do the re-balance first and then see how that goes before making the decision. The fundamental mechanics of the economic model are sound. The values which it is using presently are crazy, but that's a reason to change the values, not re-design the system from scratch. The algebra is good, it's just the math that's bad