Large language models exhibit powerful capabilities that many see as the kind of flexible and adaptive intelligence that previously only humans could exhibit. I’ll discuss three questions that arise with the advent of large language models: 1) How will large language models affect research into opportunities and problems that arise in traditional software engineering contexts? 2) What new classes of software systems will large language models enable? 3) How can we better understand the reasons behind the surprising capabilities of large language models? I’ll present projects that highlight issues that arise in each of these areas and use these projects as a starting point for a broader discussion of particularly promising directions (as well as others that are less promising).