Once the artificially intelligent machines are able to identify objects, we might want to teach them how to value different things differently based on their utility, demand, life, etc. How can we accomplish this and how did we start to value things?
From your question I can assume you are a beginner in the field of AI. Welcome to this exciting field.
To answer your question, we have not yet been able to create a truly artificially intelligent program. They are all apparently intelligent but are just a set of simple/complex rules. An artificially intelligent agent must have at-least 2 aspects in the inside of its head/program. The ability to logically derive conclusions, and a capability for learning (these combined with ability to to take inputs and respond).
Now, logical reasoning part is the field of AI. Lots of simple codes performing complex task already exists.
Your question is based on the learning part, which is handled by Machine Learning programs. They learn iteratively. Object recognition is only one part of ML. They can also predict. Anyways, object recognition is done on the basis of maximizing reward/minimizing penalty (Cost function). Now, this minimizing of penalty is done by giving different weights or valuing different attributes of the objects differently. So this has already been accomplished. Depending upon the task at hand, we supply the attributes which are related to the final task ,otherwise the attribute is of no use. Like economics has no influence on weather.
We also have this thought process by selecting things depending on the goal we need to accomplish, sometimes we select things consciously sometimes we are hardwired by genes to do so (Touching fire). So you see, ML methods are modeled on how we do things and we do things based on weight-age to different factors. If we don't give proper weight-age we either learn by the penalty/punishment imposed upon us or our genes get wiped out from existence and thus the same behavior of not giving proper weight-age is not passed on.
So in short, a ML learning algorithm recognizes objects or does some final task (especially in games like chess) based on the utility or influence of the given features of current state on the final result.
You have asked two questions. How humans began to put a valuation to things and how to accomplish the task of valuation within an artificial intelligence construct. Human valuation is accomplished through trial and error experience, subjective choice and relative comparison, among other things. Valuation in an AI construct would be data-driven, objective and perhaps absolute. The choice of a valuation method would be determined by the choice of or desire of outcome.