The AI Handicap files tells you a lot about weakness in AI; after all one does not on purpose give the AI a handicap on what's advertised as standard in things it does well.
For the versions prior to BNW, the AI was to Chieftain; basically the developers opinion was that some of the AI flavor combos mimicked behavior so bad that it was expected by a Human on Chieftain.
BNW though introduced the "AI default handicap level" that allowed them to tailor it in specific ways.
Happiness Default : 15. This field is your flat happiness count before you found any cities. That value is actually the same as that of a Human on settler; note that this is 3 more than Human on Chieftain.
So what does this tell us the developers opinion was: Some of the AIs flavors for building a worker is really, really slow and so the AI needs a big boost to give time for them to finally get around to it without crippling the economy. (Note that in the very early game it's really bad to go negative as growth crashes)
In actual practice: If the Human selects Immortal + the AI doesn't need this at all since the AI starts with a worker at the highest levels. So if you only play at those levels, you can safely mod that back to 9.
Num City Happiness Mod & Population Unhapiness mod: 90. For comparison purpose this is about half way between Human on Warlord and Human Prince. Note that it gets further multiplied if Human is on King and above
So what does this tell us the developers opinion was: Some of the AIs have low happiness / high expansion combos in which if a human were to do that on Prince would bring them to their knees; but it would be too big a bonus to the high flavor AIs to go all the way down to Warlord.
In actual practice, at Prince & King that is correct. However, this interacts badly with the multipliers, and in my own opinion if the Human only plays on Emperor and above they can mod this up to 100 and just use that level's happiness bonuses.
Route Cost Percent: 50. This basically means the AI pays half the road maintenance costs. (AI kind of is Inca even when not playing them). This is the Chieftain level.
So what does this tell us the developers opinion was: Some of the AIs build road networks that are too big / or perhaps the high AI worker flavor ones start constructing a road too early.
In actual practice: I've seen no evidence of AIs road networks being so bad as to need an ongoing bonus and would mod it away.
Unit Cost Percent / Building Cost Percent 75. This basically means the AI gets a 25% discount building anything. This is between Chieftain & Warlord. But it's likely to multiply.
So what does this tell us the developers opinion was: Some AIs building flavor are so low they need big help when they actually build one; meanwhile other AIs unit flavors are so low they need big when when they build one.
In actual practice: While this is likely to be the case on Prince, On Immortal+ stacks badly with bonuses given to AI there, so if a Human only plays on Immortal+ better to mod these away and use the per difficulty level ones there.
Research Percent: 85. This means the AI gets a 15% discount researching techs. For comparision, the value at Settler is 90. That's right, the discount is bigger than a human on Settler.
Developers opinion: Some of the AIs science flavor is so low with the new per city increase in science costs, that the AI needs major help.
In actual practice, this is unfortunately correct; to safely mod this away you'd need to set every science flavor to 10 at the same time.
Policy Percent: 75. This means the AI gets a 25% discount discovering polices. This is between Chieftain & Warlord.
So what does this tell us the developers opinion was: Some of the AI cultural flavors are so low they need big help with policy rate even without cultural victory being tied to it.
In actual practice: Probably true on Prince. But in my own experiments on Immortal the AI doesn't fall behind with this set to normal and in fact still run ahead of the human (because of other bonuses that allow AI to construct cultural buildings faster ...)
"AI Barb Bonus": This one doesn't matter because it won't be read; it's the copy on the human's that matters.
So going to Prince; human at this level is assumed to need a 40% help against barbs while the AI needs 60%
Developers opinion: AI kind of sucks dealing with barbs
In actual practice: This is true. Not recommended to change any of these lines.
AIUnitCostPercent / AIUnitSupplyPercent / AIUnitUpgradePercent :This copy doesn't matter either because it won't be read; it's the copy on the human's that matters.
So going to Prince;
AIUnitCostPercent; this is 85% on Prince while 100% on Warlord.
Developers think AI needs a bigger army to deal with bad tactics that would be expected of a Human on Warlord.
In actual practice: That is correct; I wouldn't touch this one.
AIUnitSupplyPercent: This is 20% on Prince while 0% on Chieftain.
Much like the above only this may deal more with AI having forces scattered as much as could be expected (as opposed to concentrated) of Human on Chieftain.
In actual practice: That is correct; I wouldn't touch this one.
AIUnitUpgradePercent: This is 50% even on Settler
Developers think the AI needs help upgrading their army. Might also be leftover from Civ IV and never changed as that's the value it ended up with the BTS expansion.
In actual practice: That might be true in early game, but the AI has plenty of cash come mid game when most unit upgrades are, and so this bonus could be removed.
For the versions prior to BNW, the AI was to Chieftain; basically the developers opinion was that some of the AI flavor combos mimicked behavior so bad that it was expected by a Human on Chieftain.
BNW though introduced the "AI default handicap level" that allowed them to tailor it in specific ways.
Happiness Default : 15. This field is your flat happiness count before you found any cities. That value is actually the same as that of a Human on settler; note that this is 3 more than Human on Chieftain.
So what does this tell us the developers opinion was: Some of the AIs flavors for building a worker is really, really slow and so the AI needs a big boost to give time for them to finally get around to it without crippling the economy. (Note that in the very early game it's really bad to go negative as growth crashes)
In actual practice: If the Human selects Immortal + the AI doesn't need this at all since the AI starts with a worker at the highest levels. So if you only play at those levels, you can safely mod that back to 9.
Num City Happiness Mod & Population Unhapiness mod: 90. For comparison purpose this is about half way between Human on Warlord and Human Prince. Note that it gets further multiplied if Human is on King and above
So what does this tell us the developers opinion was: Some of the AIs have low happiness / high expansion combos in which if a human were to do that on Prince would bring them to their knees; but it would be too big a bonus to the high flavor AIs to go all the way down to Warlord.
In actual practice, at Prince & King that is correct. However, this interacts badly with the multipliers, and in my own opinion if the Human only plays on Emperor and above they can mod this up to 100 and just use that level's happiness bonuses.
Route Cost Percent: 50. This basically means the AI pays half the road maintenance costs. (AI kind of is Inca even when not playing them). This is the Chieftain level.
So what does this tell us the developers opinion was: Some of the AIs build road networks that are too big / or perhaps the high AI worker flavor ones start constructing a road too early.
In actual practice: I've seen no evidence of AIs road networks being so bad as to need an ongoing bonus and would mod it away.
Unit Cost Percent / Building Cost Percent 75. This basically means the AI gets a 25% discount building anything. This is between Chieftain & Warlord. But it's likely to multiply.
So what does this tell us the developers opinion was: Some AIs building flavor are so low they need big help when they actually build one; meanwhile other AIs unit flavors are so low they need big when when they build one.
In actual practice: While this is likely to be the case on Prince, On Immortal+ stacks badly with bonuses given to AI there, so if a Human only plays on Immortal+ better to mod these away and use the per difficulty level ones there.
Research Percent: 85. This means the AI gets a 15% discount researching techs. For comparision, the value at Settler is 90. That's right, the discount is bigger than a human on Settler.
Developers opinion: Some of the AIs science flavor is so low with the new per city increase in science costs, that the AI needs major help.
In actual practice, this is unfortunately correct; to safely mod this away you'd need to set every science flavor to 10 at the same time.
Policy Percent: 75. This means the AI gets a 25% discount discovering polices. This is between Chieftain & Warlord.
So what does this tell us the developers opinion was: Some of the AI cultural flavors are so low they need big help with policy rate even without cultural victory being tied to it.
In actual practice: Probably true on Prince. But in my own experiments on Immortal the AI doesn't fall behind with this set to normal and in fact still run ahead of the human (because of other bonuses that allow AI to construct cultural buildings faster ...)
"AI Barb Bonus": This one doesn't matter because it won't be read; it's the copy on the human's that matters.
So going to Prince; human at this level is assumed to need a 40% help against barbs while the AI needs 60%
Developers opinion: AI kind of sucks dealing with barbs
In actual practice: This is true. Not recommended to change any of these lines.
AIUnitCostPercent / AIUnitSupplyPercent / AIUnitUpgradePercent :This copy doesn't matter either because it won't be read; it's the copy on the human's that matters.
So going to Prince;
AIUnitCostPercent; this is 85% on Prince while 100% on Warlord.
Developers think AI needs a bigger army to deal with bad tactics that would be expected of a Human on Warlord.
In actual practice: That is correct; I wouldn't touch this one.
AIUnitSupplyPercent: This is 20% on Prince while 0% on Chieftain.
Much like the above only this may deal more with AI having forces scattered as much as could be expected (as opposed to concentrated) of Human on Chieftain.
In actual practice: That is correct; I wouldn't touch this one.
AIUnitUpgradePercent: This is 50% even on Settler
Developers think the AI needs help upgrading their army. Might also be leftover from Civ IV and never changed as that's the value it ended up with the BTS expansion.
In actual practice: That might be true in early game, but the AI has plenty of cash come mid game when most unit upgrades are, and so this bonus could be removed.
