Is investing in deep technical knowledge still worthwhile for a software developer in the age of AI?

Is investing in deep technical knowledge still worthwhile for a software developer in the age of AI? I have been working as a software developer for about 10 years. During my career I have become proficient with many programming languages, frameworks and tools. I can use the technologies I work with effectively, but I have often felt that my understanding of them was much shallower than I would like. For years, I mostly learned things when I needed them for a project or a job interview. Before interviews, I would often have to refresh things such as design patterns, SOLID, UML or object-oriented programming principles, even though I had been working with related concepts for years. I knew how to use the technologies, but I didn't necessarily understand them deeply. Recently, I started trying to change this. I discovered that I can learn and retain much more using spaced repetition, without forgetting what I have learned. I am now trying to understand the technologies I use at a much deeper level. JavaScript is a good example. I have used it throughout most of my career, but only recently started studying the ECMAScript specification in depth. I even wrote a partial implementation of the specification as a learning exercise. I have also gone through the MDN JavaScript documentation and the YDKJS books, to cover JavaScript concepts that are not part of the ECMAScript specification. At the same time, I am wondering whether this is still a good investment of time. With AI becoming increasingly capable of writing and explaining code, I wonder how much value employers will continue to place on this kind of deep technical knowledge. Will it still be valuable to have a very deep understanding of the technologies you work with, or will the ability to efficiently use AI to produce working software become much more important? I am also concerned that I may eventually be rejected in job interviews because I have invested too much time in deep technical knowledge and not enough in whatever skills become more important in an AI-assisted development workflow.