While consumers are increasingly engaging with AI-powered tools, they are not yet ready to rely on them for the moments that matter most, particularly when money is involved, according to a published report. The findings come from an analysis published by CX Dive, which examines how major brands are deploying AI and how consumers are responding to those efforts.
At a high level, the data tells a nuanced story. AI is proving useful, but usefulness is not the same as trust. According to research cited in the report, 85% of consumers say AI-generated responses are helpful, but only 71% say they trust those responses. That gap becomes even more pronounced deeper into the customer journey, where actual transactions remain limited.
Where AI is working today
- Early-stage interactions like answering questions, providing guidance, and helping consumers navigate options
- Product or solution discovery, where stakes are lower and speed matters more than precision
- Enhancing self-service experiences without fully replacing human interaction
Consumers are clearly open to AI in these contexts. In fact, many are already using it without thinking about it, particularly in search and recommendation engines.
Where AI is struggling
- Final transaction steps, especially those involving payments or sensitive financial data
- Situations where accuracy, accountability, and trust are critical
- Open-ended environments where AI can be manipulated or “broken” by users
The report points out that consumers are still hesitant to enter payment information into AI-driven environments, especially third-party platforms. Instead, they prefer to move from AI-assisted discovery into familiar, brand-owned channels to complete transactions.
Interestingly, that behavior may actually improve performance. While traffic from third-party AI platforms to brand sites is declining, conversion rates among those who do make the transition are increasing. In other words, AI may be acting as a high-quality filter rather than a full-service channel.
Key considerations for implementation
The report emphasizes that successful AI deployments are not broad or generic. They are highly targeted.
- AI should be tied to specific use cases, not deployed as a blanket solution
- Guardrails are critical, as users will actively test and attempt to break systems
- Data control and security must be tightly managed, especially in customer-facing environments
- First-party AI experiences tend to generate more trust than third-party tools
There is also a clear warning against treating AI as a “checkbox” initiative. Implementations that lack a defined purpose or tangible value are far less likely to gain adoption.
What this means for collections and consumer engagement
For organizations trying to increase engagement rates and drive consumers toward self-service, the takeaway is not to pull back from AI, but to be more intentional about where and how it is used.
AI is most effective as an entry point. It can reduce friction, answer questions quickly, and guide consumers toward action. But when it comes to closing that action, especially payments, consumers still want the reassurance of a trusted environment, according to the report.
