The art in artificial intelligence
When you ask a layman what is artificial intelligence the most probable answer that he would give you is, ROBOTS! But AI is not just robots rather I’d say AI is an art.
But in general when some ask me what is AI?
There is not one but numerous different definitions, with different slants. However, the definition which I prefer is this: There are a number of cognitive tasks that people can do easily—often, indeed, with no conscious thoughts at all—but that is extremely hard to program on computers. Standard examples are vision, natural language understanding, and ''real-world reasoning:" Artificial intelligence, as I define it, is the study of getting computers to carry out these kinds of tasks.
Now, it can be said that this definition is inherently unfair to computers. If you define AI as "problems which are hard for computers," then it is no surprise these are hard for computers.
For the time being, though, I think it can be agreed and I believe that whether or not these abilities are necessary for a super-intelligent computer they could certainly be an asset. Therefore, in speculating on the singularity, it is somewhat relevant to consider how well these tasks can now be done, and what are the prospects for progress.
However, it is very critical to understand how far the state of the art is from human-level abilities. The point is best illustrated by the example. The following are representative of cutting-edge research in computer vision and natural language processing.
Recognizing Birds In Images. A program which recognizes main components of a bird's body (like body, head, wings, legs) and even identifies the category of bird (duck, heron, hawk, owl, songbird) in pictures of birds is said to achieve a precision of about 50 percent on finding the components, and about 40 percent on identifying the category.
Identifying Images That Match Simple Phrases. A program which was developed to identify images in a standard collection that match simple phrases. This was a very successful experiment for some phrases; e.g. at a 50 percent recall cut off, the precision was about 85 percent for "person riding bicycle" and 100 percent for "horse and rider jumping."
Coreference resolution. A state of the art system is for coreference resolution—identifying whether two phrases in a text/sentence refer to the same thing or two different things—achieved success rates ranging from 82 percent recall and 90 percent precision to 40 percent recall and 50 percent precision, which depends on the source of the text and the grammatical category involved.
Event extraction. A program for identifying events of a particular type in news articles; specifically, for identifying the event trigger, the arguments, and their role. For instance, in the sentence "Bob Cole was killed in France today," the trigger for the event die is "killed," the arguments are "Bob Cole" and "France" and the roles are victim and place respectively. There are 33 different event types.
However, the success rates for such AI tasks generally reach a plateau, often well below 100 percent, beyond which progress is extremely slow and difficult. Once such a plateau has been reached, an improvement of accuracy of 3 percent—e.g. from 60 percent to 62 percent accuracy—is noteworthy and requires months of labor, applying a half-dozen new machine learning techniques to some vast new data set, and using immense amounts of computational resources. An improvement of 5 percent is remarkable, and an improvement of 10 percent is spectacular.