Whilst the uses and potential benefits of AI-enabled technologies are expanding, there is also increased recognition of their potential downsides. This short comment critiques approaches to AI-enabled technologies based around ‘calm computing’ for rendering both users and the technologies themselves as passive and thus doing little to challenge the unsustainable status quo. It highlights some recent efforts to develop more proactive forms of AI but argues that, whilst these may lend more agency to technologies, they do little to increase the agency of users or engage them in thinking more critically about the nature of contemporary inequalities and unsustainabilities. To move beyond this situation, it calls for new forms of experimentation around what might be termed “Provocative AI” technologies. Drawing on emerging speculative examples relating to gender inequality, energy, and surveillance, it suggests that, instead of minimising or solving problems for users, Provocative AI could instead work to create more actively engaged citizens by promoting deeper reflection on the nature of contemporary societal problems.