How you study AI will influence how capable you are in solving AI related problems. There are 2 approaches to studying and understanding AI and Machine learning in general, the Math view and the Code view.
When studying machine learning and you're reading some book or watching some video, most of the examples are usually given in Math form. This is very important to understand because it is shorter and more succinct in expression. Books and papers are mostly written in math form because it makes for more economical expression of ideas that are very technical and not very expressible in natural language.
Another form of expressing ideas is in code form, and this is actually how you will deal with systems in practice. If you are well versed in math you could read some complex expression and understand everything in it, but if you are not well trained in the math discipline then reading expressions written in math can be difficult.
Everyone in machine learning should have some math, both discrete and continuous under their belt. But this should not limit you if you are still developing these skills which take some time to master.
The code view or sometimes pseudocode view is clearer, more explicit and sometimes much easier to understand in a practical sense but depending on the programming language might not be the shortest thing to include in materials like books or papers.
Although short algorithms can be written in pseudocode rather than a concrete programming language, if you are doing something like studying the code for a particular neural network architecture, the code might contain code that is actually doing the real neural network computations and other dependencies that have to be included for the sake of the requirements of the particular programming language itself.
As an AI person, you should be able to read math and implement that math in code because this is important for converting research into practical implementations.