I remember seeing a name for this, but I can't find it anywhere now.

In one of my courses, I use a weighted list to decide the choices for an AI.

Basically, each AI class has 2 standard functions:  Value and Do.

[Value] returns a float, a percentage of how imperative it believes it is that it does this.  I.e. when your base is under attack, sending troops to defend returns a high value.

[Do] executes the AI.

Then an AI Manager gets the value for each AI choice, and chooses to do the one with the highest value.

I've seen a more formal name for this, but I don't recall what it was.

Technically, when we talk about 'weighting' we usually mean some system that takes all the options into account, but factors the weights together to make a choice.

For example, if you could choose any of the options, but are more likely to choose the high-scoring ones, this is technically known as 'fitness proportional selection' or 'roulette wheel selection' in some circles, but is often just called 'weighted random': https://en.wikipedia.org/wiki/Fitness_proportionate_selection

But if you're just picking the highest-scoring option each time, that's not really a 'weight'. It's basically a score which yields a rank. The simple act of making a choice based on a score or ranking is known as a utility system, based on the game theory concept of utility as being an abstract score for how important or useful something is. Some utility systems will actually use roulette-wheel selection rather than just picking the top-scoring action.

Great!  Utility System AI was it.  I'd seen it before but trying to find the name based on the description was proving near impossible. 

 - Thanks!

http://intrinsicalgorithm.com/IAonAI/2013/02/both-my-gdc-lectures-on-utility-theory-free-on-gdc-vault/

http://intrinsicalgorithm.com/IAonAI/2015/10/building-a-better-centaur-ai-at-massive-scale-gdc-ai-summit-lecture/

http://intrinsicalgorithm.com/IAonAI/2011/12/getting-more-behavior-out-of-numbers-gdmag-article/