For me textbooks are the worst form of learning and I mostly learn by projects. For AI, build your NN framework from scratch, for statistics, find interesting datasets and perform the analyses you learn about on them. Math can be a little tougher, but my favourite part is applying it to problems I wouldn't know how to do without knowing it. For example with basic linear algebra, you could program a super basic 3d visualizer.

Also, a lot of textbooks I've encountered have a huge problem with the initial motivational part and jump straight into proving stuff which does not seem useful. Maybe try Coursera or Khan academy for the intro and then read the book after you get the basic.