Today I had my first discussion for a course on Artificial Intelligence I’m taking. In the week since the course started we really haven’t covered much material (as is to be expected) so the TA gave us a very broad topic to write about.
In short the prompt was: What is Artificial Intelligence
To suppliment we were given two definitions:
Weak AI: Machines can act intelligent.
Strong AI: Machines have minds.
My Thoughts:
In my understanding, Artificial Intelligence is the ability of machines to learn and use what it has learned to draw conclusions. As abstract as this is it still boils down to mapping inputs to outputs, which is what machines do every day. The inputs in the case being not only its perception of the current situation, but the sum of all of its experiential knowledge and the output being its action.
To accomplish this, weak AI is sufficient. Strong AI (based on the definition above) requires a machine to be something that I doubt any machine can be: non-deterministic. Having a mind and being conscious are attributes of sentience; something that humans and animals have. I do not believe that human behavior can be described as deterministic, or at the very least, the amount of inputs that are mapped to an output in a human is too difficult to quantify and infeasable to simulate in a machine.
In short, I do not think that strong AI is attainable. This, however, I do not think that this in anyway is a limitation on artificial intelligence. A turing test, though I do not think is sufficient for intelligence, is an example of this. The test consists of a person communicating with either another person or a machine. The two cannot see each other and interact only through text. If the person cannot determine whether the other entity is a person or a machine, then the machine is said to be intelligent. This can be (and currently is being) accomplished by machines that don’t have minds, conscious, or exemplify human behavior. All the machine needs to do is simulate human behavior. This can even be accomplished without intelligence. If a machine has a massive database of answers to questions and when asked a question simply looks up the answer and spits it out, is it intelligent?
If a machine can analyze a question and based on a database of its previous experience conclude an answer to that question does it have a mind?