Ways to Program an Artificial Intelligence Unit
![Picture](/uploads/1/7/7/5/17758591/1587771.png?283)
There are many approaches to programming an AI, to list a few there are: Cybernetics or brain simulation, symbolic, cognitive simulation, logic based, Anti-logic/scruffy, knowledge-based, sub-symbolic, bottom up, computational intelligence, and then there’s statistical. Each type of programming has many different theories, with many possible errors and many different ways to succeed. AI’s are the computer’s brain, so many different ways of programming are needed for the computers to function the way they are needed to. The most popular way to program an AI is the Logic-based approach, meaning that the Artificial Intelligence program does not need to “think” like a human, being much more “Mechanical” or acting more like a computer, sticking to the most efficient way of performing a task, using trial and error to find the best way.
Artificial Intelligence Theories
![Picture](/uploads/1/7/7/5/17758591/2844151.jpg)
There are many theories, one of them being Turning’s “polite convention”- saying that we don’t need a machine that can act human, we need a machine that can complete its tasks. Many people look at AI’s from a philosophical viewpoint, not seeing the real science behind it. Some even saying that we are creating the end- due to so many video games and movies using AI’s as the “problem” or cause of the story. In Gödel’s incompleteness theorem, it says that “A formal system, cannot prove all true statements,” meaning that we cannot create an AI program that can be the “godlike” thing we are trying to create, it will perform it’s task, and it will do it well. Nothing more, nothing less. In Searle’s strong AI hypothesis, he says that “The appropriately programmed computer with the right inputs and outputs would thereby have a mind in exactly the same sense human beings have minds”
B Martensen
THS
2/20/13
THS
2/20/13