What is AI?

What is Artifical Intelligence (AI)? Wikipedia call it “The term artificial intelligence is also used to describe a property of machines or programs: the intelligence that the system demonstrates. Among the traits that researchers hope machines will exhibit are reasoning, knowledge, planning, learning, communication, perception and the ability to move and manipulate objects. General intelligence (or “strong AI”) has not yet been achieved and is a long-term goal of AI research.” Is that it? Can it really be summed up like that? Lets look at what’s in the name.

Artifical

In this era, we certainly know the word artifical. One could define it is the simulation of a natural object or substance using man made processes. We have, for example, artifical sugars, and artifical limbs. So why do we want to replicant natural things? We have several compelling reasons.

• Reduce Costs
  Every wants more money. Sometimes, the cost of something like natural grass, is a lot more cost prohibitive than say, astroturf. It could be both production or maintinence costs.
• Improvements
  Some artifical objects boast improvements over their natural cousins. Artifical sweetener is one such example: Diabetic users don’t have to worry about their blood sugar.
• Replacement
  Some natural things cannot be regained. A lost limb, for example, cannot be regrown (as we know of). For this reason, it may be nessesary to replace it with a reasonable copy.

Intelligence

We had it easy with artifical. Intelligence is a word of debate. Are dog’s intelligent? Are birds? Insects? Would a definition of ‘responding to stimuli’ be correct? I don’t think so. I think the definition of ‘being able to respond to the environment based of previous learned experiences’. The big point in that is ‘learned’. Once again, learning is fuzzy. You can learn that tacks can hurt, but also learn that sharp pointy things can hurt as well.

Artifical Intelligence

Lets put a few things together now. Would it be correct to say that artifical intelligence is the acquisition of information to be used to improve or replace a natural process? I think that’s pretty bang on. But does it fit our AI definitions of today? Game AI is notorious for being dumb, but for what reason? In the whole AI scheme, I think they lack the learning part. I had an interesting discussion about a game AI with a co-worker, and that in this game, the AI always does the predictable, and doesn’t vary from this pattern, as well as doesn’t change it’s tactic based on pass losses (that’s the whole learning part). Hard wired AI just can’t work, it’s not ‘learning’. Is a chess AI learning? Probably not. It probably know’s the rules, and knows how to find the best move (again, hard wired). I challange programmers to come up with an AI that starts out not knowing anything, and learns how to beat the player.

Are you up for it?