Artificial Intelligence is a thing people study to try and make computers be intelligent or rather be able to think, learn, decide, understand, etc. the way humans do. Strong AI would be totally simulating the way a human thinks. Weak AI is more aimed at doing one particular thing similar to how humans might. AI is similar to human intelligence in that we are trying to simulate exactly that, human intelligence. What makes it different is right in the name: Artificial. This “intelligence” is not natural. We as humans can tell a computer what to do and how to make decisions. We can tell it where to look to learn new things. We can inform it how to interact with other humans. We can even make it look like a human. But none of it is natural or really real. In many cases of AI that exist today, the machine is doing what the developer tells is to do and learning what the developer tells it to learn.
Things like AlphaGo, AlphaZero, and Deep Blue seem to be more weak AI if AI at all. When “teaching” a computer how to play a game, you are teaching it how to analyze data and choose a “move” based on the data’s outcome. A machine will perform better the more games it plays, just like a human will perform better the more games they play, however the way they learn is very different. A machine does not require sleep or food. They have one job and it is to play the game for 24 hours a day, 7 days a week, and learn everything it can. It also can easily check the outcomes of certain moves from past experiences and remember winning percentages of certain moves. This last part is something a human does not do when playing a game. When a human plays a game they have to go with their instincts. They have to consider their options and trust their gut they are making the right decision. This could mean they choose a safe move over a risky move or vice versa. A human can take things into account that a computer may not consider, like if they have played the same person many times and they know they will make safer choices or if they feel really lucky that day. There are certain instincts and signals that a human can pick up on, that even math cannot predict.
Watson is a little different because it is more open ended than just playing a game. Watson was created to be able to answer questions. This seems a bit more complicated because questions on Jeopardy can cover a wide variety of topics, and now Watson is mainly used in the medical field. Watson has been trained to do many different things. This seems like more of a human-like situation. Reading about some of the errors that Watson made in his rounds of jeopardy just remind me that Watson is not human. He made mistakes that a human would not make, like answering “1920’s” after another contestant already guessed “the 20’s.” There are certain common sense things I do not think we can teach computers.
I am a fan of the Chinese Room counter argument to the Turing Test. I agree that a machine that is merely “simulating the ability to speak Chinese,” is not really understanding Chinese, and rather outputting what a developer’s algorithm thinks it should output. Can we really say a machine is intelligent like a human if it cannot fully understand what it is saying? One good counter argument to the Chinese Room test would be that the definition of artificial is “made or produced by human beings rather than occurring naturally, typically as a copy of something natural; insincere.” This would mean that Artificial intelligence is indeed simulating human intelligence, not necessarily having human intelligence. However, this is not the way that many people perceive AI.
I am more concerned about the intentions of those who create AI than I am about AI itself. The computer responds in the way the developer has told it to respond. The people behind the AI need to be thoughtful in what exactly they are telling the computer to do. For this reason I find it hard to believe that a computing system can be thought to have a mind of its own. A human can learn things from other people and be taught how to do things, but we have the ability to make our own decisions and judge if something is right or wrong, or if we want to do it. We can decide if we want to learn the things we are being taught. A computer does not have these capabilities (at least for now…).