Wedding Minister Letter

Inteligencia Artificial: Turing contra Searle

As I read Microserfs are going to do from Sunday Turing Test (1950) to see if artificial intelligence can overcome the final test, impersonate human without being recognized. It is interesting, and certainly very important technological researchers. But the Turing test was more relevant to the purely technical part, also sought to explain to humans through the machine, explaining and defining intelligence.

There are two theories that address the issue within AI itself since it also can deny both and consider that human intelligence can not be explained or defined by computer modeling.

The first theory is known as strong AI , argues that the mind is a consequence of the functional organization of the brain. Any system that reproduces the functional brain organization possess identical emotional states and intentions conscious experiences. If it looks smart is smart. Thus, if the product of an interaction not recognize that we are interacting with a machine, then it is intelligent. The problem is that there is very little subjectivity and interrater reliability in this.

If now I go to a gas station, pick up the hose and the machine tells me "Good morning , you have chosen super petrol ", fill the tank, then put the hose in place and the machine tells me" Thank you for using our services. goodbye, "then I must conclude that the machine really think. And more, it also feels thanks. If it were not already know the operation could come to believe that the machine actually knows that I arrived, I wish me good morning and thank use their service. Fortunately, we know how machines work well enough to know that this is not true, and enough to not behave like the typical villager image that comes to town, and when the machine tells "Good morning" he responds "Hello, how are you?" , and angry with the machine not to respond to your question. We say, somewhat Turing was impressed by the technological evolution of the moment and some naive sinned reaching this conclusion, as that "village in the big city."

Anyway, there is something much better to understand this, and do the test yourself. There are lots of programs that simulate the interaction with other individuals through chat *. Myself I did one in his first day on I put on the list. I was quite surprised that I hit a piece as it is translated though, and instead put my name "Turing" because in my opinion gives some humor to the thing to imagine him saying that:

(...) Turing: These are two that go on a bike and falls in the middle. Computer
: Clearly you're not human.
Turing: Yes I am, you are the damn machine.
Computer: No, I'm a real person.
Turing: No, I am the real person. Are you crazy or what?
Computer: Yes, I'm crazy.
Turing: So if you are a computer, what computers you can go crazy?
Computer: I have also emotions, like a human being!
Turing: I knew I was right!

undeniably have their true complexity. "Talking" with Dr. Abuse, also even learn something, I have to talk about how difficult it was Spain win the World Cup. One seeks to find the fault and the topic you know nothing and sometimes surprised. It is also true that sometimes have systems go unnoticed if they have memorized answer, but it is easier to brew than it seems.

To counter this theory appeared exaggerated idea of weak AI , which believes that only when the machine simulates intelligence serves as a means of understanding how human cognition, but the computer itself has no inner experience or self-consciousness, or intentionality, but simply a useful descriptive model of operation. It was Searle who the metaphor of "The Chinese Room" (1980) contradicted Turing's position. According to Searle, thinking the machine has states of mind and intelligence is wrong. I, not to repeat the example of the Chinese Room [which, incidentally, no longer identical to the existing programs translators] otherwise I shall, and so you have both. Suppose we are now going to Japan and worked as a doorman at a hotel, give us a paper with these instructions:

-When entering a customer say: Konnichi wa
"When you give a tip di: Arigato
"When they leave the hotel di: Sayonara
" When you make a mistake say: sumimasen

saying it could spend weeks without knowing what it is, and therefore, not feel, however complex and extensive than was the list actions and responses that give us (which is what would happen to these programs simulation). Yet anyone who heard you would think you really know Japanese (whatever that interact with the machines might think they are smart). But would you thank in that language because they tell you when you get a tip to say something (indeed, not even tell you what you say when you feel gratitude, I say they say when they give you a tip, feel what you feel). Also is that what you said is due , but could have been a hazing and you, naive, with a smile say something rude to the customer, you do not know what you say, you're just "programmed" to do so, as the machine the gas station. Computers in the Turing test, therefore, is to feel and have mental states simply are prepared to give a series of specific answers.


And is that even this would be correct from the point of view of many. Because it does not mean that because something works give the same answers like (feel or not feel) or serve to explain anything. You can reach the same conclusion through different mechanisms.

* Make yourself the Turing Test ( programs botchatter ):

• Turinghub (English)
• Jabberwacky (emotions can teach and use indicators) (English)
• Dr. Abuse (computerized psychotherapist) (English)
• iGod (simulated chat with God) (English)