Ask HN: Future proofing software engineering skills against AI

So whilst I don't think my job as a software engineer / backend server dev is about to be taken over by AI, it's pretty clear that AI tools will be able to, for example, build CRUD apps in the not too distant future. Not sure of the timescales but if this isn't starting to really take off in 10 years I'd be suprised. Obviously one solution is to move up the stack and get better at guiding the AI, writing the specifications, data models or whatever inputs might be required by this, as yet, hypothetical AI system. High level systems thinking and mapping that to the messy problem domain will become more of a focus as the lower level detail wlll be increasingly automatable.

What skills, other than this higher level work do we see sticking around? Understanding security and infra generally, i.e. moving down the stack for an app developer also seems like a good option. As well as reasoning about the software behaviour we need to ensure security, reliability at a human level to some degree. It just seems like writing app code for many standard use cases will be a solved problem by some AI system so staying in the middle long term, might be a dead end?