I met Andrew Glassner at #SIGGRAPH back in the early 1990s, when I was still a #ComputerGraphics academic researcher. I remember him as a superb communicator. His recent book is a non-#CG one: "Deep Learning: A Visual Approach" (2021). This book contains no equations at all, only simple diagrams and illustrations. A tour de force of visual communication, that. Go, CG!
https://www.amazon.com/Deep-Learning-Approach-Andrew-Glassner/dp/1718500726
I like #AI as a computation tool but, as my friends on MathsTodon know, I genuinely detest the common way #DL is now being taught to #ComputerScience and #DataScience undies—no maths, just API call-fest. Consequently, new CS and DS grads entering #IT—who are fashioning themselves as "#AI experts", either due to profit motive or through sheer ignorance—lack the adequate mathematical and intuitive understanding of this rather "deep" subject. Andrew's new book will not remedy their maths shortcoming, of course, but it surely will give them a thorough, intuitive grasp of DL, IT's current "darling of the day".
Indeed, this book is so gentle that I dare say 8th graders could be taught DL, out of it. That is my ardent endorsement of this book.
Armed with this intuition, those youngsters with adequate background in undergraduate maths could move up to books like "#DeepLearning", Goodfellow (2016) and "Pattern Recognition and #MachineLearning", Bishop (2006).