I want to help "reach" AI. The whole notion of "AI" itself is a topic for discussion, but let's just say my notion of AI is along the lines of what was discussed in Turing's "Computing Machinery and Intelligence" paper.
I believe attention is a key component for that. Thus, in order to "reach" AI, we need to "reach" attention. We first need to formulate a precise and useful definition for "attention". BEFORE THAT, let me cite some examples of the phenomenon I'm trying to describe.
Examples of 'attention':
For me, all these events are related. This seems to be a very important function for us. We have finite "resources" to think and we must use it wisely to survive! In the next session, I try to define it more precisely and from a computer science standpoint.
Towards a more precise definition:
In a world with scarce resources, attention is about orchestrating processes.
Data is virtually infinite.
The number of programs are also virtually infinite.
Thus, the number of processes, that is, the manipulation of some data by following the set of rules of a program, is also infinite.
However, the quantity of resources to execute these processes is finite.
Attention is the act of allocating resources to processes.
In other words, attention is the entity that, given context and goals, selects a set of processes (program-data pairs) and determines the quantity of resources to execute each of them.
SO, my questions to you all are:
What do you think about it all? Do you think it makes sense? Do you think attention can be thought as a phenomenon in itself? Do you think this definition I lay out is useful for us to help guide work on the subject?