Do people interact mindlessly or mindfully with in-car AI?

Wang, Luo and Liu conducted in-depth interviews with 17 owners of Nomi, the voice-based in-car robot built into NIO electric vehicles, whose "humanoid face-like interface with oval eyes" swivels and blinks to address each occupant. The authors report that "most participants perceive Nomi as a human-like actor and interact with it mindfully," naming it, dressing it in clothes and hats, and "projecting emotions onto it." Only "a small group denies its social status while still holding social expectations for it." Most participants describe Nomi's social role as "a combination of a pet and a friend" and see it as "an independent entity rather than one that is reliant on humans." The authors frame this as support for the Media Evocation paradigm (van der Goot & Etzrodt, 2023), which treats intelligent machines as "philosophical provocateurs" that prompt reflection on human-machine ontological boundaries, superseding the Media Equation / CASA framework's assumption that social responses to computers are automatic and unreflective (mindlessness).

The mechanism the authors give is cue quality: "Highquality social cues, such as voice and facial expressions, can trigger mindful anthropomorphic responses" with gaze and facial feedback making owners "perceive communication as bidirectional." This leads the authors to argue that the criteria separating mindfulness from mindlessness have become blurred: under the original CASA standard, mindlessness means responding socially to cues while explicitly denying the machine's social status (Kim & Sundar, 2012; Nass & Moon, 2000), but Nomi owners increasingly do not deny that status — their "natural responses are shaped by accumulated experiences from the era of human-machine interaction" rather than by deliberate, moment-to-moment reflection, yet the attribution of agency persists regardless. The authors explicitly position this against Nass and Moon's original view that training people toward mindfulness "tends to complicate the process and is neither typical nor common" — this study finds the opposite: conscious recognition of a machine's social nature has become "a widely shared understanding."

This sits close to Do more social cues always make AI feel more present?, which already establishes that primary cues like voice and appearance can trigger either mindless or mindful anthropomorphism; the Nomi case is a field case that this specific population leans heavily toward the mindful branch rather than splitting evenly, and that the mindful/mindless boundary itself is harder to draw than the original MASA operationalization assumes. It also extends Do humans apply human-human scripts to AI interactions?: Nomi owners' scripted, non-deliberate responses (dressing the robot, treating it as a pet-friend hybrid) still carry explicit agency attribution, suggesting that script-driven and reflective engagement are not mutually exclusive categories in practice, as the sharper mindful/mindless split implies. The dressing-up and naming behavior also instantiates the anthropomorphic-features and autonomous-action hallmarks described in What design features make users perceive AI as conscious?.

The study is a single-product, single-market case (17 owners of one Chinese EV's voice assistant), using interview self-report rather than behavioral measurement, so it cannot establish that mindful anthropomorphism is becoming the norm for AI interaction generally — only that it predominates in this sample of people who chose to personalize and talk about their experience with Nomi. The authors themselves flag that individual factors (gender, personality, prior usage) and contextual factors (family dynamics, usage setting) were not examined and could change the mindful/mindless balance. The finding supports, at most, the narrower claim that high-quality, bidirectional-feeling social cues in an embodied in-car agent can shift self-reported engagement toward conscious, agency-attributing anthropomorphism rather than denied, automatic responding.