Well the main cost of units is that they cost gold per turn. Furthermore early units like chariots or axeman arent exactly much more costly hammerwise anyway. I prefer even a chariot to have as a barbarian defender. If you are past the early stage cost of hammers really doesnt matter so much at all.
This is so painfully untrue that I don't know what to say

. GPT is the lesser consideration relative to hammers. Unless you're talking unit supply early on, but that's not dependent on unit type and therefore irrelevant. Just spawnbust the barbs...usually by the time you attack they've slowed down anyway.
Of course the less the computer cheats the better. For such a complex programm the KI seem to do a very good job without cheating that much - at least you wont notice it too much. That said the KI has to cheat at some things to have a chance to win against experienced players and thats just not avoidable.
Using exploits is just a different story. One doesnt need to use that to have a chance to win. That said I dont know the wall chop thing we are talking about, it just did sound like a typical exploit to me. If its a feature it would be nice, however how can a chop trick be related to a trait (like its better used for protectiv) ?
I agree they did a pretty good job given complexity, but the AI is still utterly terrible. For it to compete reasonably with competent humans, we have to spot it all kinds of handicap bonuses. However, agreed-upon handicaps are within the rules set at game start. The AI cheats to know things that we do not agree upon:
- Lying about disposition in some situations
- Map hacking for trades
- Map hacking in terms of the ability to see the max unit move distance for all its units
- Detecting BTS power irrespective of espionage
Among others. They ARE material and you can notice them easily (I lost a game not too long ago because the AI detects worst enemy trades before it knows you, conveniently making me worst enemy upon contact....does that sound normal to you?!).
Also, you use the term exploit quite loosely. Diplo manipulation offers a far greater advantage than wallchop, and can be done with any trait just by knowing AI tendencies. The same people who define a wide range of tactics have no trouble memorizing the patterns of a given AI...but the latter is something not available to AIs, and provides a much greater advantage.
Wallchop is best with PRO due to its discount on walls (100%), making whips and chops 100% more effective. It's possible to get more overflow gold this way, since overflow is capped at the amount the building costs. Keep in mind, however, that OVERFLOW ----> GOLD WAS DELIBERATELY ADDED INTO THE GAME, it used to just be wasted. It also does not = win and frequently is not even your best use of those trees! Labeling such a thing as an exploit is kind of a bad joke, IMO. Similar to how people claim choking is exploitative even though doing it can utterly screw you. If a maneuver offers a valid cost/benefit that is not obvious, it does not provide an overwhelming advantage, and it was put into the game based on its general mechanics...how the hell can anyone define this as an exploit?
I guess the apostolic palace and UN wins are also exploits? That's why I hate the term. There's a difference between valid tactics and cheating. Exploits in a gaming sense are problematic to begin with...what truly separates an exploit from a strong tactic? How "realistic" it is? Balance? What argument are you using? Wallchops certainly don't make PRO a top tier trait even if you use them.