artificial intelligence assumptions

[Isaac]

I'm trying to comprehend what artificial intelligence is. It's a difficult topic for me, but I'd rather know something than nothing at all.

So far, when I see coding examples I either don't understand what I'm looking at or it's nothing other than a series of "if" statements.

Assumption 1: An AI is nothing more that a loop that tests conditions, like a series of "if" statements and reacts to them or does nothing.

Assumption 2: From what I understand, a learning AI would be able to write new conditions as it's running.

I assume I'm overlooking something more complex. Can you help me figure out what that is?

[Jeff250]

How does a human brain work?

[Isaac]

I don't know, but simulating an organic neural network, especially a human one, isn't something most examples are doing. In fact it's ground breaking. I'm not pretending I can write a speech recognition program or a self driving car.

What the majority of AIs are doing appear to be more narrow in capabilities.

Can you give me any feedback about my assumptions in the original post?

[Jeff250]

The term AI is currently used very broadly. In some problem domains, like writing an artificial opponent for tic-tac-toe, it might be a series of if's or rules. If you write a neural network that can tell the difference between a photo of a cat and a dog 80% of the time, many people would call that AI as well. I don't think there are any generally accepted rules for what AI must be though, although for some problem domains like conversation, the Turing test is often used as a benchmark. For many people the ultimate goal of AI is to make Lt. Commander Data, i.e., something that is human, so it might be something that somehow combines the techniques of a good tic-tac-toe opponent, something that can discern the difference between cats and dogs, and a fluent conversationalist, among other things.