Research on automated software engineering aims to improve software development by automating tedious or error prone manual activities. We can now automatically produce tests, models, patches – and even software itself. But can we ultimately remove the human from software engineering, and is this even desirable? In this talk, I exemplify in the domain of automated software testing that software engineering depends on human software engineers, and discuss implications this has on automation research. Automated techniques need to be developed for software engineers, not to replace them. We can, however, utilise both, explicit and implicit human intelligence, to take automated software engineering to the next level.