Answer 1

That is very much interlinked, as machine learning is the foundation of predictive model building, a cornerstone of artificial intelligence.

Answer 2

ML

Answer 3

The first thing to understand is that AI is a goal, and ML and DL (deep learning) are the means. 
ML and DL provide models and frameworks to achieve AI. 
First, understand the high level goals of AI.
Then, equip yourself with the means by focusing in ML and then DL.
DL involves complex models that are built on top of ML concepts. Hence, first ML and then DL.
Then, learn how the ML and DL techniques are used to achieve NLP goals.
Revisit AI which uses a lot of ML, DL, and NLP.

Answer 4

OK I answer directly in short that MACHINE LANGUAGE

Answer 5

You can implement machine learning in any language really, I would probably get a good grasp on python if you aren't currently a strong programmer and want a starting point, otherwise just start learning about machine learning, the concepts and algorithms don't depend on the language, just your implementation will. I took an AI class in university that used Java personally. You might also find courses that often give code samples in a certain language, if you see one come up a lot, just start learning it so you can keep up.

Also, looking at the tags on your post, take a second to read what each language you've heard of is used for, some like java/c++/python are vaguely equivalent and practical for making AI, but you'd only ever use HTML or CSS if you're making AI that generated webpages or something like that, since those two aren't programming languages.