Should AI question users or just recommend decisions?

Parker and colleagues (Australian Journal of Management, 2025) open their argument with a scenario from Tim Miller, "the professor of artificial intelligence who developed this scenario (Miller, 2023)": two colleagues give advice on a crucial decision. "Bluster" tells you what he thinks you should do "at length, and then justifies why this is the case." "Prudence" instead "asks you what you think you should do, and then provides feedback, with evidence to support the decision and evidence against it," revising her feedback as your view changes. The authors note "for many purposes you are likely to prefer Prudence," yet argue "the predominant design of contemporary AI is akin to Bluster: AI provides a recommendation and explains why the AI system made that decision."

The reasoning for favoring Prudence-like design, which the authors attribute to Le et al. (2024), is that "the human worker maintains control of the decision, leveraging their human expertise, and the AI generates clearer bounds of accountability, reduces over-reliance on AI systems, mitigates skills erosion, and develops decision-making skills for novices." The shift is from explanation-only transparency to AI as "interactive tools driven by the human user, rather than relegating the human user to a role of oversight only," so that technology "truly enable[s] augmentation of human ability" instead of reducing people to "monitoring technical systems." Elsewhere the same authors note that explanation by itself does not solve the problem Bluster represents: citing Schömbs et al. (2024), "even when the risks and uncertainty about an AI-assisted decision systems are communicated, users tended to over-rely on and over-trust AI systems" — so a Bluster-style system that explains its recommendation can still produce over-reliance, which is part of the case for a Prudence-style design instead.

This design contrast sharpens the control-and-oversight concerns already in the library. What UX principles do workplace users want in AI agents? finds business users rank human control highest among workplace agent design principles; Bluster-versus-Prudence supplies a concrete interaction pattern — an AI that questions and evidences rather than recommends and justifies — for what control-preserving design could look like. It also contrasts with How should humans and AI systems divide decision-making tasks?'s Type A/B/C typology, which sorts decisions by where accuracy advantage already lies: Parker and colleagues instead argue for redesigning the interaction itself, independent of which party is more accurate, so the human stays the decision-maker. And it offers a design-level response to Does incremental AI replacement erode human influence over society?'s worry about eroding human participation: a Prudence-style AI is proposed as a way to keep humans substantively "in the loop" rather than merely nominally supervising a system that decides for them.

The Bluster/Prudence contrast is a hypothetical scenario used to motivate the argument, not a tested interface or an empirical comparison of the two designs' outcomes; the cited benefits of Prudence-like systems — control, accountability, reduced over-reliance, skills development — are attributed to Le et al. (2024) rather than demonstrated by Parker and colleagues themselves. The excerpt does not address how a Prudence-like system would work for tasks where the human has no prior view for the AI to interrogate, or what it would cost in the speed and scale the authors concede Bluster-like systems offer.