Instead of scripting the AI to do shi-, we would instead provide the AI with a set of "win conditions" and "values" and have it play against itself and re-evaluate its strategies based on its results.
Obviously not for this game, but for future games, it's definitely something to look at. Why are humans good at strategy? In part, because we know what to ignore. And focus instead on the relevant details.
The AI in this open-ended game is probably the best you can get with scripting. I.e meh.