Realistically, at any given time there are 3 or 4 books that I'm dedicating meaningful cycles to and expect to finish "soon'ish".
Right now the ones I'm seriously working on are:
Superintelligence - I've heard so much about this book and keep hearing people talk about the dangers of AI, and while I already have an opinion on the subject, I thought it would be interesting to read what Bostrom had to say.
Abductive Inference Models for Diagnostic Problem-Solving - an older book on an approach to automated Abductive Inference called "Parsimonious Covering Theory". I'm not just "reading" the book, as in reading it straight through like a novel, I'm actually working on re-implementing PCT using a more modern software stack, with a goal of doing some research into possible ways to use abductive inference in conjunction with other techniques (neural networks, reinforcement learning, graph-based knowledge-representation, etc.)
Artificial Intelligence: A Modern Approach - such a class in the field, I felt like it was time to finally sit down and the read the whole book, cover to cover.
[0]: https://www.goodreads.com/user/show/33942804-phillip-rhodes