Discussions of the problematic relationship between AI and society have recently only been heightened. These discussions, nevertheless, remain partial until they take into account how we live AI technologies in the unremarkable circumstances of our everyday affairs. In arguing for the importance of such a noticing, this article centers on the internet-of-things and associated digital voice assistants (DVAs). These commercial social robots, designed as conversation-oriented devices, manifest their incompleteness in their need for other voices. Paying attention to that relationality at an embodied scale of analysis brings up our involvement in situated interactional production, while also manifesting its reciprocal character. This not only questions the conviction of DVA designers that these gadgets generate effects of presence in relation to an intentional mind, but it also resists a parallel return to the individual that more often transpires in the discussions of the problematic relationship between AI and society. We practice this resistance by evoking efforts in distributed cognition and the extended mind hypothesis, but we also go beyond the instrumentalist reasoning that primarily recognizes the world as carved into tools that can extend our cognition. We thereby contest the rationalization of our use of these technologies in terms of efficiency and convenience and propose a close noticing of situated engagement with voiced AI as a mode of intervening.