It doesn't matter what strategies you program into the AI if its evaluation system prioritizes ALL OF THE WRONG THINGS.  For example, if a chess computer thinks a queen is more valuable than a knight, it will lose every single time because it will try to trade its queen for knights.  

1.  Raw production is worth a lot more than anything else in the game.  

2.  Influence is pretty much worthless.  

3.  Population and rate of population growth are as valuable as raw production.  

4.  Techs that give %bonus to growth, production, research, or economy are much more valuable than other techs. Needs to be prioritized early REGARDLESS of the civilization.

5.  "Scripted Strategies" should only begin on turn 80 or so.  Early game EXPANSION and BUILD UP strategy should roughly be the SAME for ALL AI.  In other words, every AI should follow only ONE scripted strategy (the best one) in the early game.  It differentiates between strategies later.  

Teach the AI to prioritize these things and it will improve massively.