Marketers Trust AI Advice, Not AI Acting on Its Own

Marketers have made peace with AI making suggestions. They have not made peace with AI acting on its own, and a new StackAdapt survey of nearly 700 brand and agency marketers puts a number on exactly where that comfort runs out. Ninety percent of respondents are fine with AI recommending an action for a human to review. Comfort drops only slightly, to 89%, when AI prepares an action for approval, and holds near 78% when AI is allowed to act within rules a person already set. It collapses to roughly half when AI is asked to act autonomously, a nearly thirty-point drop from the top of the ladder to the bottom rung.

The gap shows up again once tools are actually in front of marketers rather than described in the abstract: across the same research, only 6% of marketers say they act on in-platform AI recommendations almost always, with the largest share dismissing suggestions as too generic or irrelevant to their specific campaign context, and a smaller group citing a lack of transparency into how the recommendation was generated. StackAdapt frames this as a “delegation gap” between what marketing organizations say they want from AI and what they are actually willing to let it decide, and the survey ties it directly to accountability: a meaningful share of respondents report no clear owner for a poor AI-driven decision when one happens.

The original insight here is not that marketers distrust AI in general, since the acceptance numbers at the recommendation end are high. It is that trust decays specifically with autonomy, not with AI capability, which means the fix is organizational rather than technical: define who owns an AI-driven decision and what happens when it is wrong, before expanding what the AI is allowed to do on its own. Vendors selling “autonomous” campaign management into this market should expect procurement conversations to focus less on model quality and more on exactly this kind of governance and rollback plan.

Source: StackAdapt