# How to define artificial intelligence (AI)? [closed]

It is very strange. Most people see AI as digital entity of consciousness (that may sooner or later turn against people - like in Matrix or Star Trek: The Movie, or even Transcendence).

But is it only one meaning of AI?

Games are (more or less) AI too - because they have some mechanisms that decide what to do next, mostly as answer to player's actions. Truth is, that these mechanisms become (sooner or later) very clear/readable for forecasting. To paraphrase one person from Simutrans international forum, It is artificial, but is it intelligent?

So, where AI begins? Does it begin with creation of consciousness - or is it anything that is capable to do simple decision what to do next (for example simple regulators - that check something and keep it on some value with opening or closing of any valve or so)?

## closed as off-topic by Mołot, L.Dutch♦, Hohmannfan, Frostfyre, AzuaronMar 29 '17 at 13:28

This question appears to be off-topic. The users who voted to close gave this specific reason:

• "This question does not appear to be about worldbuilding, within the scope defined in the help center." – Mołot, L.Dutch, Hohmannfan, Frostfyre, Azuaron
If this question can be reworded to fit the rules in the help center, please edit the question.

• Hello. This is a philosophy problem we are coping in real world, and, at least for now, it is very context and opinion based. – Mołot Mar 29 '17 at 9:49
• May I recommend AI.SE and Philosophy.SE for this question? I don't see a WorldBuilding question here right now. What are you trying to achieve? For more open discussions on a technical perspective AI.SE would be perfect and for a humanitarian discussions about the implications of consciousness Philosophy.SE would be perfect. – Secespitus Mar 29 '17 at 10:26

A dog has intelligence, so using human consciousness as a marker of AI is flawed - this is why the Turing test is based on ego, not a pure definition of intelligence.

I made a YouTube video on this very subject, I'll try to summarize.

Remember - An intelligence can remember things it has experienced

Reason - An intelligence can draw conclusions based on past experiences

Reflect - An intelligence can look back on its actions and change its behaviour based on previous outcomes

I think these three R's represent pure intelligence, and that what most people are trying to emulate is AC - Artificial Consciousness.

Realistically, they're aiming for AHC - Artificial Human Consciousness, which is a flawed target due to the many external biological influences that impact how the brain works, but it is what it is.

I've known someone who was doing his PhD in artificial intelligence in Colorado, more than 10 years ago. Aside from cool stuff like cars recognizing obstacles (which is now actually on the roads!) and prototype fly brains, he talked to me about the workings behind it all.

The most important part of what I got from him, was the learning matrix. It comes down to AI having a learning system, which at first fills in slowly and has to be spoon-fed, but eventually will have made enough connections to be able to learn by itself.

How it goes from there, doesn't really matter. Consciousness and the likes are a different discussion altogether. Being able to learn and make new connections independently, seems to be the core of AI.

I think the term Artificial Intelligence is widely used... and misused. The problem is even though everybody can agree (at least it is less discussed) on what is "Artificial" : not natural, human-made, which requires skills to make; a lot of people are struggling at defining the "Intelligence".

Anyway, the definition I think people should know is the one defined in 2003 by Russel and Norvig. To sum up, they state that an intelligence must "think" or "act" and it can do it in 2 ways "like a human" or "rationally". Thus, it gives you 4 cases in which your question has different answers.

• Think like a human : [The automation of] activities that we associate with human thinking, activities such as decision-making, problem-solving, learning...(Bellmann,1978). Here comes the problem of conscientiousness, you do not want your AI to be able to compute thinks quickly you want it to react like a human, to be aware of itself and maybe to feel emotions.
• Act like a human : The study of how to make computers do things at which, at the moment, people are better (Rich and Knight, 1991). Learn, move real objects, know how to behave in society... Here you are making an android (I prefer to say "anthropoid" but nevermind).
• Think rationally : The study of mental faculties through the use of computational models (Charniak & McDermott,1985) Here you want to put your AI in Wallstreet, and demonstrate the mathematical theories.
• Act rationaly : Artificial Intelligence (AI), broadly (and somewhat circularly) defined, is concerned with intelligent behavior in artifacts.(Nilsson, 1998) Act autonomously by taking into account its environment to fulfill a goal. Self-driving cars, Goo winners and most of nowadays robots in fact.

Now, by putting your examples in those category I think you can see the objectives of an AI, and place them on a success scale. And even if the think/act barrier is thin, the human/rational one is huge. You don't want your self-driving car to stop because it is afraid of the dark (like a human could be) and you don't want your robot assistant to bring you a saw when you say you want to "cut some corners".