Consumers are bringing their own AI tools into customer service conversations to research policies, prepare questions and challenge answers.
A Capacity survey found that 35% of respondents have used an AI agent to handle a service request, rising to 49% among Gen Z and millennials.
Many still want human control, and conflicting information highlights the need to check AI-generated answers.
Consumers seeking a refund or disputing a bill increasingly have a new assistant: their own artificial intelligence tool. While more businesses are deploying AI in customer service interactions, so are consumers.
A survey of 1,000 U.S. consumers released by Capacity, which sells customer service automation technology, found that 55% have used AI to help with a customer service request. A smaller share, 35%, have used an AI agent to handle a request at least once.
That assistance can include explaining a retailer’s return policy, researching a billing problem or preparing questions before contacting a representative, according to the company’s accompanying report.
How they’re using AI
The findings suggest consumers are using AI to strengthen their position in conversations with businesses. About 34% said AI prompts them to ask more specific questions, while 23% said it would lead them to request an explanation or citation of company policy. Another 20% said they would escalate a problem more quickly when using AI.
Capacity’s report also found that 58% would be somewhat or much more likely to push back after a business denied their request if they were using AI.
However, greater confidence does not establish that the information is correct. Some 44% reported encountering AI information that contradicted a company representative or policy. That finding alone does not show which side was right.
It’s not foolproof
Capacity acknowledged that AI can misinterpret policies, draw on outdated material or produce incorrect answers. When information conflicted, 14% said they would trust AI outright—the same percentage who would trust the representative outright.
Consumers also remain cautious about surrendering control. Only 14% would trust an AI agent to complete an entire interaction. Another 21% would let it reach a human representative before taking over themselves, while 26% would not delegate any part of the exchange.
For consumers, a practical approach is to ask AI for links to the policies behind its answers, then check those policies against the purchase date, receipt and terms of the transaction. AI can help organize a complaint, but a persuasive explanation still needs supporting evidence.