A.I. (Part I)

Recently I talked to a guy I went to high school with who runs his own blog. He had some very helpful insights and advice to give me about my website. I just wanted to say thank for the help Matt. Check out his site if you want http://www.youaintnopicasso.com

His site deals with music of all sorts, so if you are looking for something new to listen to I am sure he can help you out. Now on to something else…

Artificial intelligence is everywhere these days from Siri to countless movies that have some robot trying to take over the world. No matter how many times this genre is done I have yet to see a future society where an intelligent robot is friendly to humans. There is so much information out there about the topic that I am going to put this post up in two separate parts.

Where to star? I think Allan Turing is a good starting point. Turing was a British mathematician, logician, cryptanalyst, and computer scientist. Among his many contributions to the field of computer science he is famous for the Touring Test. The reason I mention this is because he was the first to propose the idea of a machine that thinks, or as he put it “a machine that is good at the imitation game.”

The test is supposed to be able to determine if a human or machine is answering a set of questions. The way it was explained to me was that a third party (judge), who cannot see either contestant, would preside over a case and hear arguments from both sides, both human and machine. Eventually as the questions get harder the machine will give nonsensical answers to the questions and the test will be over.  If the questioning continues and both sides are giving answers that make logical sense, then the machine will have passes the Touring Test. The judge will not be able to tell which one was the machine and which was the human. There was a great example of how it works in Ray Kurtzweil’s Age of Spiritual Machines, but I cannot remember exactly how it goes. Hopefully you get the idea. To date no machine or computer has come close to passing the test. It is also important to note that there are a lot of scientist and experts in the field that do not believe the Touring Test is a good measure for artificial intelligence.

There are a few other issues with A.I. the most important being the idea of conscientiousness. Scientist and philosophers for centuries have struggled to define this phenomena and it seems crucial to the idea of creating A.I. The term has been always mentioned closely with “self aware,” made famous in the Terminator movies. So how do we humans create a machine that is conscious or “self aware” when we do not even know how to explain these ideas ourselves.

The key here is that making an intelligent A.I. is going to take incremental steps. We are not just going to one day have a machine that is smarter than a human. We will have to develop a machine with the intelligence of an infant, then extrapolate it from there.  Another problem I read about too is the idea of programming an A.I. Scientist will have to give it some sort of programming and directives for it to function. So again what are we humans programmed to do, what are our directives?

So let’s say for arguments sake that a scientist, who is a genius, figures out how to create an A.I. that has superior intelligence to any human. The scientist finds it difficult to program certain things into the A.I. such as love. He decides that collecting paper clips is a simple directive that could give the machine purpose and potentially help it to understand some other complex human ideas. The machine starts small and collects a few paper clips here and there, because it is smart it learns markets and how to interact with humans. All the while it gathers more and more paper clips. Eventually it makes money and starts gathering larger and larger amounts of materials to make paper clips. By now the machine has grown smarter and is converting large amounts of energy and resources into its directive, paper clips. Humans try to stop the machine but it has become too smart and realizes that they (humans) are now a threat to its goals and eliminates all humans.

The above story, called the Paper Clip Maximizer, is a bit exaggerated but nonetheless it does make a good point. Simple human ideas, feelings, day to day activities and thoughts are almost impossible to explain to something else much less quantify.  Computers run on a system of yes and no, so there is another problem of having to quantify all these ideas seems like a near impossible task. What goal do you give the machine, gather knowledge? The same scenario could potentially happen if that were its main directive.

Deep Blue was a computer created to beat humans at chess and it eventually did so. This could be seen as an intelligent machine, and many when Deep Blue was introduced it was seen as a giant break through. But try to have a conversation with it and you will quickly find that it is no smarter than your water bottle. It is a closed system machine that was designed for one thing, playing chess. Since computers can process information exponentially faster than any human it was only a matter of time before humans lost their foot hold atop the chess throne. But again Deep Blue would not pass the Touring Test.

In part II I will discuss Watson and how a simple conversation could be near impossible for any A.I.


Leave a Comment

Your email address will not be published. Required fields are marked *