AI is transforming software development at an unprecedented pace. Organizations are delivering software faster than ever, but speed alone is a misleading measure of success. Every technological revolution creates hidden costs, and AI is no exception. Behind the productivity gains, a new generation of risks is emerging: AI-generated technical debt, software degradation, governance debt, loss of maintainability, and growing dependence on opaque AI ecosystems. Left unmanaged, these hidden costs may outweigh the very benefits that made AI attractive in the first place. In this talk, Valentina Lenarduzzi and Davide Taibi present their research vision for the next generation of software engineering, where success is no longer measured only by how quickly AI generates code, but by how well organizations engineer software that remains trustworthy, maintainable, and sustainable over time. Drawing on their latest research, they discuss the emerging challenges that will define software engineering over the next five years and outline the new methods, metrics, and governance approaches needed to build AI-powered systems that organizations can trust for decades, not just deploy in days.