Evolutionary and genetic programming is AI for dummies, you could start experimenting with that with no formal training and make progress.
Neural nets will probably get you stuck fast though, you better take classes to understand what the hell you're doing.
Name:
Anonymous2012-09-08 14:05
Most commonly recommended AI book seems to be Artificial Intelligence: a modern approach. It's pretty good, though it covers much more than genetic algs and neural nets.
For languages, the ones with the most Machine Learning libraries seem to be Java, Matlab, R, and Python. I actually couldn't find terribly much for C++. Dunno about LISP or Haskell.
Name:
Anonymous2012-09-08 15:08
I second Artificial Intelligence: a modern approach. I'd like to add that AI is all about logic and statistics. Ensure that you have a firm grasp in both fields. I'm assuming you already read your SICP.
Name:
Anonymous2012-09-08 19:40
For total FIOC noobs, read Programming Collective Intelligence, it's somewhat funny.
But if you want real AI, Artificial Intelligence: a modern approach and that's it.
Name:
Anonymous2012-09-09 2:29
nobody uses neural networks or lisp/haskell.
AI has been split into machine learning, computer vision, natural language process, data mining and so on.
It's highly mathematical, so I recommend you get used to hitting the books every night. Measure theory, matrix factorization, signal processing, and so on will be your basic tools. The more math you know the better.
You will also want to learn about distributed computing, bash, and linux toolchain.
Without presentation, your knowledge is worthless, so you should know some basics of html/css/javascript.
Python,R,c++ are good to know. Traditional computer science is being merged with information science at cornell and a few other schools. The future in computing is helping people understand information they otherwise could not.