What drives shoppers to accept autonomous AI assistants?

Frank, Folwarczny, and Otterbring (Psychology & Marketing, 2025) report a five-study program — one real-world behavioral field experiment plus four controlled online experiments with "almost 2000 US and UK shoppers" — testing what drives consumer acceptance of highly autonomous AI shopping assistants. The headline finding: perceived benefits, not consumers' sense of personal power, are the "central driver of adoption intentions of highly autonomous AI in scarcity-related contexts" (Studies 3A-3B). A Google Ads campaign (Study 1) found "higher click-through rates for high-autonomy (vs. low-autonomy) AI shopping assistants" — a behavioral measure, not a stated preference. This note draws only on the publisher's abstract, so these are the authors' own summary claims rather than full study detail.

The mechanism the authors propose runs through scarcity. Across Studies 2A-3B, "scarcity amplified the perceived advantages of high-autonomy AI" — when shoppers face limited stock, tight deadlines, or competing bidders, an assistant that acts with more independence looks more valuable because it promises to resolve the scarcity faster or more effectively than a lower-autonomy tool. That amplified benefit perception is what then predicts adoption intent. The authors frame this as a reorientation of the acceptance literature: prior explanations centered on consumers' "personal sense of power" (an agentic, control-based account), and this program shifts the explanation toward perceived benefits as "pivotal drivers of AI acceptance" instead.

This sits usefully against notes already in the library. Why does AI trust keep falling even though it matters most? identifies trust as acceptance's strongest predictor at a macro, cross-domain survey level; this paper narrows to a specific context (online shopping under scarcity) and finds perceived benefit, not trust or power, doing the predictive work — a reminder that acceptance drivers vary by context rather than resolving to one universal factor. It also bears on Does machine agency exist on a spectrum rather than binary?: where Rammert's spectrum describes degrees of machine agency structurally, this study supplies an empirical finding about which degree (high autonomy) shoppers actually prefer, and the condition (scarcity) under which that preference strengthens. And it qualifies Does the personal assistant model actually serve most users?: adoption intent here rises with benefit framing specifically in scarcity-salient shopping, not from a broad appetite for autonomous assistants generally.

The abstract does not report effect sizes, the demographic makeup of the "almost 2000" shoppers beyond US/UK, or whether the benefit-over-power finding held across all five studies or only the scarcity-specific ones (3A-B). It also does not specify what "high-autonomy" meant operationally — full purchase authority versus recommendation-only, for instance — or whether the Study 1 click-throughs converted into completed purchases. The findings license a narrower claim than "people want autonomous shopping agents": they support "in scarcity-salient shopping contexts, marketing an autonomous assistant's benefits increases adoption intent," which is the managerial recommendation the authors state explicitly, not a general claim about AI acceptance.