U.S. Customers' Experiences with and Opinions of AI in Customer Service

A comprehensive ContactBabel study of 1,000 U.S. customers and 200 U.S. organizations revealed a complex landscape of attitudes toward artificial intelligence in customer service. While they acknowledge certain advantages of AI, particularly speed, they harbor significant concerns about accuracy, accessibility to human agents and the overall quality of service.

Among survey respondents who had knowingly experienced AI customer service (84 percent of the total panel), the comparative assessment reveals notable strengths and weaknesses. The greatest strength is the speed advantage. Not unexpectedly, customers generally recognize that AI delivers faster service than human agents, meeting their expectations for immediate and speedy interactions.

Among the critical weaknesses are the following:

• 61 percent report AI's understanding of issues is poorer than human agents;
• 49 percent state overall resolution quality is worse;
• 40 percent find AI provides less accurate information (perhaps the most concerning finding); and
• 37 percent perceive AI as less friendly in tone.

The accuracy gap presents a fundamental challenge. If AI is to gain widespread acceptance, response quality must improve dramatically. Poor experiences create an inoculation effect; customers become skeptical of all AI systems, regardless of how effective individual implementations might be. Industry-wide trust depends on consistent quality, as customers remember the 1 percent of wrong answers rather than the 99 percent of correct ones.

When Customers Prefer AI

Despite concerns, 66 percent of respondents identified at least one situation where they would prefer AI over human agents. They included the following:

• Quick answers to simple questions (most popular) – aligns with current chatbot deployments;
• Avoiding call queue wait times – addresses a major pain point in customer experience;
• 24/7 availability – particularly valued by younger demographics; and
• Preference for non-human interaction – important for certain personality types and circumstances.

Following up on ongoing issues showed minimal preference for AI, suggesting customers still see AI as best suited for simple, transactional queries rather than complex, contextual problems.

There were also a number of demographic variations. Younger customers, for example, show greater openness to AI across most scenarios, but even the 65+ age group recognizes AI's value for simple self-service tasks. Older customers, having experienced frustrating phone queues, are more willing to try AI alternatives. However, around half of the 55+ demographic state they would always prefer human agents. More affluent respondents demonstrate higher AI acceptance, particularly for after-hours contact. Lower-income groups show slightly stronger preference for human interaction, though 60 percent still have situations where they would choose AI.

The research also quantified six key concerns, each resonating with a majority of respondents:

1. Difficulty Reaching Human Agents (63 percent concerned). The most prevalent fear is that AI will create barriers to human contact when needed. This concern stems from real experiences with self-service systems that seem designed to prevent, rather than facilitate, agent access. Many customers have learned to game interactive voice response systems by pressing zero or selecting wrong options just to reach a person. The 65+ age group shows 80 percent concern about this issue. The critical implication here is that any AI implementation must provide clear, easy pathways to human agents. Forcing customers to remain in self-service channels is self-defeating; they will find workarounds or abandon the company entirely.
2. Inaccurate Information and Misunderstanding (64 percent concerned). Customers across all age groups fear AI will misunderstand their queries or provide incorrect information. Stories of AI hallucinations have created widespread anxiety. This concern is particularly strong among older cohorts but affects all demographics. The critical implication is that organizations should focus on perfecting simple interactions rather than attempting to solve every query through AI. Success requires near-perfect accuracy; even 1 percent errors will disproportionately damage trust.
3. Customer Service Job Losses (59 percent concerned). Contrary to the narrative that contact centers are unpopular 21st Century sweatshops, customers express significant concern about employment impacts. This likely reflects practical understanding that live agents deliver better outcomes for complex issues, combined with broader societal concerns about unemployment.
4. Data Security (56 percent concerned). Data security concerns peak among 55-64 year-olds, who are both technically aware and cognizant of past breaches. Younger customers, despite more online activity, show less concern, possibly reflecting different risk perceptions or specific AI digital literacy (or that this age group has less historic, health-related and financial personal data to lose).
5. Lack of Empathy and Understanding (55 percent concerned). While customers don't necessarily expect AI to match human empathy, they worry about reduced understanding. This concern is compounded by experiences with keyword-based systems rather than natural language processing. Customers face uncertainty; should they use simple keywords or detailed explanations? This industry-wide challenge won't be resolved quickly.
6. Hidden AI Use (54 percent concerned). Many customers feel uneasy about undisclosed AI interactions. Research increasingly supports transparency, showing that open AI identification builds trust and reputational integrity. While non-disclosure might boost short-term metrics like conversion rates, customers who later discover automation might feel deceived, damaging long-term relationships. Regulatory and ethical pressures favor clear disclosure, particularly for sensitive data or consequential decisions.

Strategic Implications

The research reveals a customer base that is often willing to engage with AI (if it benefits them, of course) but which requires specific conditions:

Customer Expectations:

• Fast, accurate responses to simple queries;
• Clear escalation paths to human agents;
• Transparency about AI use;
• Natural language capability rather than keyword matching; and
• Reliable information quality.

Organizational Imperatives:

• Prioritize accuracy over breadth of capability;
• Maintain accessible human agent channels;
• Build customer confidence through transparent implementation;
• Recognize that poor industry-wide AI experiences create universal skepticism; and
• Focus on use cases where AI demonstrably excels: simple questions, queue avoidance, 24/7 availability.

U.S. customers demonstrate pragmatic attitudes toward AI in customer service. They recognize its potential benefits, particularly in speed and availability, but demand reliability, accuracy, and maintained access to human support. Success requires organizations to implement AI where it genuinely serves customer needs, maintain transparency, and resist the temptation to use automation as a barrier to human contact, despite the cost savings being dangled in front of them.

The challenge is both technical and reputational. Individual organizations must deliver quality AI experiences while the industry collectively avoids creating widespread customer resistance. With around two-thirds of customers concerned about accuracy and agent accessibility, the margin for error is small. Organizations that respect these concerns while delivering genuinely helpful AI experiences will gain competitive advantage in customer satisfaction and operational efficiency.

Steve Morrell is managing director of ContactBabel.