Use whatever language you're best at - what language you use is really the least of your problems. I've used Prolog, Lisp, Basic, Visual Basic, C, C++, and Python - all depending on the environment I was working in, the system I had available, etc.
My advice is to read-up and learn to implement the basic algorithms - trees, perceptrons, Genetic Algorithms.
Next, work on some of the canonical problems and get a feel for how to use the tools.
After that, come up with (small) problems of your own, and work out a solution.
Google is your friend - you can find a lot of good academic papers to get you started. If you have no math and/or no coding ability, you're starting with a handicap. This can be overcome. In some ways, lack of experience might actually be helpful.
One thing you'll find as you read up on the subject is that there's plenty of introductory material, and plenty of advanced material, but very little in between.
The most important thing to learn in my experience as an amateur AI developer is how to acurately and concisely describe the problem to be solved.