I just heard a "machine learning is taking over everything" lament earlier this evening - not from ad tech specifically, even. Although I can't comment on my personal experience(being younger myself) one pathway to longer term job security for software seems to be to gradually move "down the stack" towards increasingly deep, low-level problems, where the necessary competence and experience is high enough to eliminate most young competitors. Another is to grow entrepreneurial enough to do consulting, or if more ambitious, product. A third is to transition towards an adjacent role where your specializations become more adaptable and less commodified - for me, I've pushed for a mix of design and management skills.