AI is becoming a standard part of how people shop, but there is a clear limit to the extent to which consumers are willing to abandon it. New data from Exploding Topics shows that 77.6% of consumers have used AI to facilitate their purchases in the last six months, and more than 43% do so every week.
This level of adoption suggests that AI is now integrated into everyday shopping behavior. At the same time, most of these users are not ready to let AI take the final step and complete a purchase.
This gap between usage and trust indicates the emergence of a ceiling in the AI business. Consumers are comfortable with AI to inform their decisions, but not to act on their behalf.
This divide is visible in the way AI is used today. Most buyers use it for product research and price comparison, not transactions, which keeps it firmly in the discovery phase.
Understanding the Confidence Ceiling
AI is proving effective in shaping what people buy. Around 68.64% of users say it influenced a purchase they otherwise wouldn’t have made, giving it real weight in the decision-making process.
But trust drops sharply when money comes into play. More than half of consumers are uncomfortable with AI storing their card details, and the most common amount they would allow AI to spend autonomously is $0.
Even among frequent users, most cap this figure at $50 or less. This suggests that the problem is not familiarity with AI, but trust in handing over control.
This creates a structural divide in the purchasing journey. AI becomes a decision layer that shapes consideration, but the transactional layer still depends on human action.
What this means for marketers
For martech teams, the implications are immediate. If AI influences decisions but does not close deals, then visibility into AI-generated responses becomes essential.
Nearly half of buyers start with AI or use it to validate their choices, meaning being cited or recommended can directly impact conversion paths. This is where generative engine optimization starts to play a bigger role alongside traditional SEO.
At the same time, perception becomes a constraint. Only a small portion of consumers believe that AI-powered shopping tools primarily serve them, while many believe they instead benefit platforms and advertisers.
This skepticism carries over to new features like AI-driven payment. Even regular users describe these tools as suspicious or unreliable, slowing the adoption of fully automated trading.
The result is a two-speed market. AI adoption will continue to grow, but its role will expand unevenly across the funnel, with influence extending beyond execution.
The takeaway is simple. AI can guide decisions and shape demand, but for now, the final step still lies with the consumer.
A summary of the report is available here. (No registration is required)
