I have lately realized that Artificial Intelligence can be best achieved if it is modeled on democracy. Democracy works in a decentralized manner and all the decisions it makes go through rigorous debate. The best of all, there is no single entity responsible for decision; hence no scope of blaming things to one entity.
The AB (simple single celled type) will be composed of several modules with different priorities/weights. Say, module ‘motion’, ‘hunger’, ‘vision’, etc. When vision sees a food it gives the analysis of trajectory to food. Then motion estimates the energy required to grab up the food and vision estimates amount of energy in the food. And hunger tells if the endeavor will be worthwhile.
Something like this:
Hunger says Yes
Motion says No, path too difficult
Vision says Yes, a tasty pizza ahead
They would decide among themselves about the final decision: if AB will spend energy to fetch food. Each modules vote their consent for a proposal made by module X. Then, taking into the votes and the weight age attached to voting modules, X decides if to go ahead with the proposal. In a nutshell, AB will work like a democracy. I guess our brains work like democracy too!
Therefore, an AB will consist of N number of modules each with a weightage for other modules. Therefore there will be N(N-1) weights for an AB.
Now, to get the ideal values of weightage we can put a lot of ABs in a common environment to compete for a limited resource such as energy. The best ones survive and reproduce; worst ones perish.