As AI is increasingly used to automate development tasks, the center of gravity in software engineering is shifting from coding to specifying intent through natural language requirements. In this talk, I explore the evolving role of requirements engineering (RE) in an era of AI-driven development. Drawing on Jackson and Zave’s classic formulation of the relationship between requirements, specifications, and domain assumptions, I argue that the fundamental challenges of RE remain more important than ever, particularly given the increasing risks of implicit specifications and assumptions embedded in AI-generated code. I further argue that RE skills are poised to become first-class competencies for software engineers, and outline potential research problems and opportunities in this space.