The emergence of the ‘new AI’ has led some to suggest that the tasks of HCI have been either solved with the use of, for example, AI-enabled natural speech interaction between person and machine, or, if not, has created a new but narrow domain for HCI research - in the area of ‘explainable AI’. But I think both these views radically misunderstand the nature of AI and the role that HCI needs to have in making AI-related applications better and more usable. In this talk I will show, with examples from both consumer-oriented AI applications and AI in more arcane scientific domains, how the role of HCI is more important than ever and turns around canonical problems of rendering the functioning of computer systems and processes in relevant and accountable ways. I will suggest that the AI community has failed to recognise or comprehend these and the HCI community itself has not been robust enough in articulating their importance. In short, my talk is a clarion call for a confident, assertive HCI in the age AI.