Artificial Intelligence: Can Computers be Human?
Artificial Intelligence is a very murky subject, mainly because the subject is very new. Although researchers have studied it since the 50s, it is clear that the subject is far more advanced than they once believed. What the researchers found out was that we don't even know enough about the human mind to begin to understand artificial intelligence. Not only are there so many functions of the brain but also many evolutionary traits that allow brains to outperform computers. Will computers ever become as powerful or even better than humans?
Artificial Intelligence is the attempt of computers to mimic or outperform the human brain. The test is that can it do a large number of the human brain's functions. Sometimes we simplify AI and even our own human brains assuming that everything can be mimicked so easily. However, if you think it through you have to imagine how many processes your brain is running and how intelligent it really is.
When Deep Blue II defeated Garry Kasparov in chess, people assumed it meant a new era of computers being smarter than Kasparov. However, this is absurd. The Deep Blue II is probably one of the dumbest pieces of software alive, it simply calculates all the possible moves, and during the matches with Kasparov, it was re-initialized with different strategies based on the "operators" who decided for the machine which strategy it should use to defeat Kasparov in "this round." In fact, the match wasn't fair, as Kasparov's games were all studied by the developers and Deep Blue II, while Kasparov was not given any logs or previous games of Deep Blue II. In fact, Kasparov demanded the logs from IBM because he believed it wasn't fair that the developers were modifying the Deep Blue II between matches.
The complication of AI can be imagined. For example, the simple definition of a word can have a vast number of meanings, and then there are slang terms that use it improperly; how can your brain tell which one is correct and which one is not? Now imagine the computer having to differentiate between thousands of other words and trying to understand the rules in order to decide whether a meaning is incorrect or correct. Even our own brains are confused most of the time.
There are four requirements for a machine to be human-like. Four requirements that differentiate humans from machines. One is human emotion. The second requirement is the ability to create it's own data associations in order to make decisions and differentiate between true and false. The third requirement is self-consciousness, which involves identity, a track of time, and a sense of the world around you. We have 5 senses and a sense of time, and a sense of who we think we are or might become. The fourth requirement is creativity and imagination, something that will probably be the last thing to be implemented properly.
One thing we must understand is human emotion. Some people do not understand why we even have emotions; some even try to hide their emotions most of the time. The reality is, emotion is an evolutionary trait that makes us have a purpose in life. Without emotion, we would not reproduce, complete our daily tasks, or interact with other humans.
The reason why there hasn't been much progress in artificial intelligence yet, is because the subject does not just need programming knowledge but vast information about psychology, sociology, philosophy, computer engineering and computer science.
An artificially intelligent machine would still only be a simple calculator without emotion. It would have to drive them to do certain tasks and to accomplish goals, which we do everyday. Scruffy Symbolic AI is really a calculator, it simply is told the concepts and uses them without learning it.
Ability to Create Data Associations and Decide
Data association is what's important with AI. Computer programs need to find ways to adapt and collect it's own data, and associate this data with other data. Most likely many of these things will be accomplished once natural language and communication is understood by AI developers.
There are Natural Language AIs out there, but none of them are close enough to actually have a real intelligent conversation with humans. Even if it came close, it still doesn't feel anything. One such Natural Language AI exists at a-i.com:
However, it is simply responding to what I say to it, rather than acting like a person. It's pretty much like the paper clip in Word.
Data association is probably best accomplished if AI can read the internet, make millions of associations, and accurately differentiate between true and false statements and come up with a decision. It also needs to be able to re-study its own associations, and "change its mind" with more information. Data sorting is a huge problem for AI. It needs to also be able to come up with it's own properties for just about anything.
Creativity & Imagination
Of course, something that has puzzled many is where to come up with new ideas, new strategies, and imagine new things. This must a be a separate algorithm using context and combining with imagination and then evaluating the creation and retrying. One company is trying to build a creativity machine. This is probably the hardest things to create. However, I think creativity is based on things we've already know and seen, then minor variations on those ideas, and since everything is very related anything is possible.
Can AI exist without these Requirements?
Without the requirements, the AI would only be a simple database much like Google, only mimicking what humans expect from it. It would still be like your calculator giving you the answers. It's called a Scruffy Symbolic AI. Some AI developers are trying to create non-symbolic AI.
The whole AI field is filled with researchers working on several aspects of AI: reasoning, planning, learning, communication, perception, and manipulation. All these processes are very complicated and may take decades to come up with really intelligent machines.
In Pop Culture
Many movies and TV series have talked about the subject of AI and how the future will be. The movie A.I. portrayed human emotion being added to a robot child and how it would interact with humans and the problems that come with it. The movie I, Robot portrayed one robot that had human emotions and an unrealistic plot of hurting humanity. Terminator, one of the best known AI/Robot related movies in history, discussed an AI program that figured out that the only way to protect the world is to take the humans out of the world and declaring war on them and even sent terminators back in time to kill the future human resistance. All were very entertaining, most very unrealistic (if Skynet was smart enough to declare war, it would be smart enough to be reasoned with), but also very realistic in other aspects.
Terminator: Sarah Connor Chronicles, is a wonderful TV series based on the movies, where it explores the idea of a female terminator (named Cameron), that might possibly develop human emotions and even possibly feelings of love for the main character. Also, whether robots would relate with each other or would they form factions against each other as well, some even opposing "skynet," the villain of the show.
However, computers must be built to develop emotions, it cannot just "invent" them itself, unless it has some ability to reprogram itself. Humans have this ability through mutation and evolution. For generations different emotions developed, making humans more adaptable and survive for more generations. Humanity is the result of millions of years of the genius of nature.
We should not be so quick to assume we can replicate such amounts of work in such a short time. Of course, ethical problems must be thought of as well, too much control, leading to bugs, can result in catastrophic effects. There was a computer programmer who had built the software for a nuclear power plant, and the plant had a malfunction, nothing happened but a meltdown would have been catastrophic. Testing is not enough for such complicated programs.
Humans will continue to have a very important role in the controlling of the world. Machines will continue to have a role of "computational" or "calculation" work. Machines need to develop their own ability to categorize information, create properties for concepts, and understand the world around them; these are very complicated processes.