For the past 80 years, “programming” meant translating a high-level, semi-formal specification of a desired effect from natural language into computer instructions, using an artificial language. Mastering these translations requires domain knowledge of algorithms and data structures, talent, and years of practice. Large Language Models (LLMs) are disrupting the very notion of “programming.” The disruption is profound, and at two levels: (1) LLMs are capable of doing those translations automatically, and (2) many of the desired effects can be obtained without the use of algorithms or data structures. This talk is an exploratory tour through this brave new world, and its consequences to our field and to CS education.