The subject is a word game. A very annoying one that when finally figured out, you realise we don’t know enough to utilize it in critical areas of human society and neither does AI.

So let’s walk this down, an artificial intelligence is in terms of intelligence simply something that can compare or combine different filters on large sets of data. We call this deep learning which then in turn means using neural networks to sort out complex problems. Which is what i mean with different filters. A analysis system using different layers to get in depth and specific contextually relevant information.

The way a machine makes decisions is unknown to the public(so they say) or is a underdeveloped area. Even though it is in my view most important that we learn something from the how rather than just accepting temporary magic to make decisions for you.

Still there is some things already quite clear when it comes to how it works, to me at least. An AI black box system makes decisions based on a scoping mechanism, where it uses general formules to create a way to answer many similar questions. A filtering of detail that ensures it can answer without changing the approach.

Like a module that can receive to multiple variables and produces answers without changing the base format of that formula.

As an example, an AI can produce a simple answer to 29 + 12 as 41, when you ask how it came to this answer it would give you the answer as how you learned it in school. so for example 9+2 carry the 1 and add that to the other order of 10’s and you get 41. This might be correct, but it is not what is going in in the black box.

Something like a series of calculations that sort of approximate the right answer and then in the end it gets some kind of half right number and corrects that because it needs to end in 1 and 41 is then the only close answer.

On itself, interesting to learn from.