What is Artificial Intelligence?

[JTC]

Please bear with me...

In a curriculum in mechanical engineering, there is a set of standard topics: statics, dynamics, vibrations, fluid mechanics, thermondynamics, etc.

What is Artificial Intelligence, and what does it entail?

(at the moment, it seems like a buzz phrase that does not conjure to me the rigor of a discipline like mechanical engineering.)

Forgive me for appearing disrespectful to those in the field, but to me it seems like a world of glorified it-then logical tests in a massive computer program. I want to understand where I am off the mark.

I hear people are coding in A.I. using various languages. But I coded in the area of mechanical engineering and used a lot of if-then tests.

What makes A.I. so special, so distinct? Is it just a buzz word for "big computer program that can monitor financial transactions?"

Yes, I know it contains "computer vision" but to me, that is sort of like "edge detection and analysis."

And that becomes if-tests and logic.

Or does the term encompass special types of computer languages more suited to include "thinking"

But then, what is "thinking?"

Do AI codes include fuzzy logic, or attempts to include "learning?"

But what is "learning" other than to "code a database of mistakes to avoid?"

I am serious... what is A.I.?

 

[ISamson]

AI is a field of study:
https://en.wikipedia.org/wiki/Artificial_intelligence
Numerous articles on artificial intelligence:
https://theconversation.com/au/topics/artificial-intelligence-90
Dictionary

artificial intelligence
noun

1. the theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages.

This is also outstanding:
http://www.abc.net.au/news/2017-08-07/explainer-what-is-artificial-intelligence/8771632

Artificial intelligence has jumped from sci-fi movie plots into mainstream news headlines in just a couple of years.

What is artificial intelligence?
AI is a computer system that can do tasks that humans need intelligence to do.

 

[256bits]

Forgive me for appearing disrespectful to those in the field, but to me it seems like a world of glorified it-then logical tests in a massive computer program. I want to understand where I am off the mark.

Hi JTC.
Actually, that is one way of putting it.
I would hate to be the person, though, or team, responsible to make an attempt to program what is being done with the present day AI only using if-then-else statements.

So there are neural nets, which simulate how actual neurons function, genetic algorithms which simulate how genes mutate as concepts for programming a machine AI. Certainly, down at the gritty machine code, binary code, boolean logic applies.

Eliza was one of the first AI's computer programmed, where the user would interact with the machine and the machine would interact back, in its best way possible to emulate a conversation.
The way the Wiki write up description does kind of give credence to your 'glorified' - preaching to the choir, as it is so over the top on trying to describe something which is rather simple in construction.
https://en.wikipedia.org/wiki/ELIZA
You can look up other articles that give a more useful description.

https://www.codeproject.com/articles/13136/chatterbot-eliza
has a c++ version and you can see it is just a bunch of if-then-else statements picking out keywords from the user and formulating a response.

Neural nets and genetic algorithms are more sophisticated, as they have the ability to "learn".
For example, you know a human face when you see one, and you are able to put names to certain faces of people you know.
To program a machine that same feature, is daunting, and yet here we are with Apple using face recognition for user identification.
All in a very small package.

Gotta go.

 

[Rive]

Forgive me for appearing disrespectful to those in the field, but to me it seems like a world of glorified it-then logical tests in a massive computer program. I want to understand where I am off the mark.

Guess you already dug up the depths of google and can already see that under the term of 'AI' there are a lot of not-exactly-matching definitions, as if actually nobody knew what's it all about.

My private definition is a kind of extrapolation based on this observation. An adaptive system is an AI if it is known what kind of expectations it should adapt to - but nobody actually knows the exact way that how will it do that...

Well, I can't really expect this to be helpful

 

[JTC]

OK, so with these responses, it seems that it is just coding. Advanced coding, yes, but coding.

 

[donpacino]

What makes A.I. so special, so distinct? Is it just a buzz word for "big computer program that can monitor financial transactions?"

Yes, I know it contains "computer vision" but to me, that is sort of like "edge detection and analysis."

And that becomes if-tests and logic.

Or does the term encompass special types of computer languages more suited to include "thinking"

But then, what is "thinking?"

Do AI codes include fuzzy logic, or attempts to include "learning?"

But what is "learning" other than to "code a database of mistakes to avoid?"

AI is more or less a buzz word. Its a catch all field for things relating to making something "more intelligent" for lack of a better term.

Edge detection and analysis can be a tool used to implement some level of artificial intelligence. Same with fuzzy logic or neural networks.

 

[DavidSnider]

All computer programs reduce to IF-THEN-ELSE statements at some level. That doesn't diminish them, that's just their building blocks.

AI operates at a level much higher than that.

A common misconception is that "computer programs can only do what they are told". Sure, the boundaries of what it can do are made by the programmer, but within those boundaries they can do all sorts of things the programmer never really imagined.

If you program a computer to reprogram itself as it receives data then its behavior is no longer solely driven by the author, but based on its own experiences.

 

[DavidSnider]

OK, so with these responses, it seems that it is just coding. Advanced coding, yes, but coding.

What else would it be? This is like saying that human beings are "just genes, advanced genes, but genes".

 

[JTC]

What else would it be? This is like saying that human beings are "just genes, advanced genes, but genes".

thank you

 

[JTC]

All computer programs reduce to IF-THEN-ELSE statements at some level. That doesn't diminish them, that's just their building blocks.

AI operates at a level much higher than that.

A common misconception is that "computer programs can only do what they are told". Sure, the boundaries of what it can do are made by the programmer, but within those boundaries they can do all sorts of things the programmer never really imagined.

If you program a computer to reprogram itself as it receives data then its behavior is no longer solely driven by the author, but based on its own experiences.

thank you

 

[yahastu]

OK, so with these responses, it seems that it is just coding. Advanced coding, yes, but coding.

As a professional in the industry, I can tell you that 99% of the AI right now is not so much coding as it is "training" pre-made AI models, which sometimes requires a bit of scripting to gather various set of data (images, text, etc) and format them into the form expected by the pre-made AI model. For the past several years, the vast majority of scientific papers in this field have just been people grabbing the low-hanging fruit of taking the same model and simply applying it to a different dataset and saying, "look, ma! it works on this, too!"

Then there is a very small percentage of people actually working on ways to fundamentally improve the models that everyone else uses, and I have much more academic respect for them.

Really none of this has anything to do with anything I would call "intelligence" or "thought" or "consciousness," however. Back when the term was coined, researchers were naive enough to think that with a few years of mucking about they would figure out how to program a "thinking computer" and they optimistically named the field "artificial intelligence." Those attempts were all a complete flop. "Machine learning" is a more apt term for the modern state of AI.