MTsardakas

Sounds interesting, I've emailed you!

In my field (cinema/theatrical analytics), the value of AI/ML/Data Science is quite easy to calculate as we mainly use models that produce results that are then checked by human experts. To produce the same amount of output of just a single model, we would need 10x headcount.

I do agree though that an estimation of the value of AI work should be the first step in a project - often it's better to ask "why do we need this" even before asking "can this be done?".

It depends! If you want to see unusual examples of applied maths or a bit of thinking out of the box, focusing on techniques and not theory, start with Howison's. Logan builds theory alongside applications and is, therefore, heavier on maths. If you're confident with graduate level physics though you should be fine with both.

Howison's Practical Applied Mathematics has many illuminating examples of maths being applied in a variety of fields. A good introduction to modelling techniques with a nice balance between theory and examples is Applied Mathematics by Logan.