It makes sense to learn AI and its mathematical basis (statistics and linear algebra) or unless I want to get a degree on the topic I’m wasting my time? If the answer is yes what are the best online resources/books which I can use?
To clarify I don’t want to specialize there and work as a “AI/data science specialist” (for which I believe a degree it’s necessary) as some graduates in shady boot camps expects to do. I’m just asking myself if I can teach myself enough IA to enrich personal projects in which IA makes sense and to become apre flexible professional.

Ok, first thing first: AI isn't something you *learn*, AI is the end product which makes decisions based on input data. That data can be consumed either in its raw form (e.g. a decision-tree AI) or through some machine learning model (e.g. an AI to remove malformed products based on camera input). So what kind of AI do you have in mind? Most AI doesn't use any machine learning or data-sciency methods, just regular algorithms, heuristics, and some probability. That's very general knowledge which has many uses besides your pet projects.

Probability and uncertainty, linear algebra, optimisation problems, the theory behind various machine-learning models... if you're not going to go all in on these, don't waste your time. No one will ever ask you to provide a confidence interval for your pet project data model or implement said model from scratch. You can drive a car without learning how each part of the engine fits in.

That being said, here's some literature I found useful:
- KP Murphy (2012) Machine Learning: a Probabilistic Perspective, The MIT Press
- T Hastie, R Tibshirani, and J Friedman (2016) The Elements of Statistical Learning, Springer
- David Poole, Linear algebra, a modern introduction, Thomson, 2006
- Roger A. Horn and Charles R. Johnson, Matrix Analysis, Cambridge, 2006
- James Stewart: Calculus, early transcendentals, 2016
- Jagannathan, Krishna. 2015. “Probability Foundation for Electrical Engineers”
- Imre Pólik, Tamás Terlaky: Interior Point Methods for Nonlinear Optimization
- Nocedal J. and Wright S., Numerical Optimization, Springer-Verlag New York, 2006
- Boyd, S. and Vandenberghe, L., Convex Optimization, Cambridge University Press, 2004.

@hitko thank you very much for taking the time to write such an exhaustive answer. I ignorantly used AI as a synonym of machine learning when AI includes much more techniques.
Everything considered it’s better for me to start just by getting my feet wet integrating TensorFlow models it in a couple of my pet projects and then if I start to get more curious about this world I’ll do a more in depth study of the various types of AI and of the mathematics behind them.

You'd better be learning some computer algorithms and architecture design instead.

I can recommend to use the app Anki and search on thier website for the words machine learning and ai. You find there packs with hundrets or thousands of cards that includes all definitions you need plus excercises.