Is difficult to find a precise or unique definition for this field that have been used in so many ways and so many contexts. Even there is no a unique answer, I would like to go back in history and take the original approach about this field and get some concepts from back in the day. One of the definitions that I personally like comes from Nils J. Nilsson from his book Problem-Solving Methods in Artificial Intelligence published in 1971. In his Book he wrote:
“The goal of work in AI is to build machines that perform tasks normally requiring human intelligence. Thus, prominent topics in the field include automatic methods for solving problems, translating languages, proving theorems, and recognizing visual objects. Although many of these tasks are exceedingly difficult, several computer programs have been written whose performance begins to match that of humans”.
There you have it, this would be my best answer to the question What is AI?
I would like to elaborate more about what is AI in the next paragraphs.
In Nilsson’s book it was mentioned that AI consists in computer programs that up to some point they exhibit some ways of thinking process as humans think, having memory and learning as key characteristics of such computer programs.
The term AI was coined by John McCarthy at the Dartmouth Conference in 1956 when the first running AI program was shown: the Logic Theorist (LT) written by Allen Newell, J.C. Shaw and Herbert Simon.
There has been a lot of controversy about this phrase even today many people shudder at the phrase “Artificial Intelligence” as they imagine something with human attributes this phrase, they say, overtones irrelevant emotional responses, eg., in people who think it’s sacrilegious to try to imitate the human brain 
I’m not totally convinced that Artificial Intelligence is a good name for this interdisciplinary field but I guess nothing can be done now. Artificial intelligence is well known for good or bad.
Other short answer for the question What is AI?: “AI is the field that makes computer models that have the ability to learn or deal with situations". Since we define intelligence as the ability to learn or understand or to deal with new or trying situations therefore computer programs with these characteristics must be intelligent. There's a great paper by Turing who really came up with this idea of AI as creating models of human behavior