(New York, NY) Even as artificial intelligence (AI) tools grow more sophisticated, cardholders still want a human on the line when it matters most. Auriemma Group’s latest issue of Mobile Pay Tracker reveals that despite the rise of chatbots and generative AI (GenAI) tools like ChatGPT, live agents remain the most used customer service channel, especially for complex or sensitive issues like fraud, disputes, and multi-step problem resolution.
“In moments that require trust, empathy, or judgment, cardholders continue to rely on people—not bots,” says Jonathan O’Connor, Senior Manager of Research at Auriemma Group. “Digital tools offer convenience, but trust is still earned through human connection. Issuers who balance automation with meaningful service support will be best equipped to meet cardholder expectations.”
Interactions with live agents—whether by phone (40%), chatting by website or app (32%), or in-person (27%)—outpaced AI-based communications within the past 12-months. Only 21% utilized a chatbot or virtual assistant, and even fewer interacted via AI-based text messaging or with an interactive voice response (IVR) system over the phone (13% each). While automation can streamline simple tasks, the findings show that cardholders still lean more heavily on human support.
This distinction is especially evident in support preferences. Even as digital tools expand, cardholders still favor human support for nearly all service interactions, especially those involving complexity, risk, or disputes. Nearly 9-in-10 cardholders prefer human support for multi-step issues and to identify or resolve potential fraud. Automated support is more welcome for simpler tasks, like resetting passwords or replacing damaged cards.
“AI is not an absolute replacement for human support,” says O’Connor. “While many issuers are exploring how GenAI—and agentic AI in particular—can transform their servicing capabilities, they must caution against removing too much of the human element. When trust and empathy matter, cardholders still want to speak to a person.”
While some (22%) believe AI-based customer service solutions improve overall service quality, nearly twice as many (41%) say they reduce it—signaling a potential backlash if providers over-automate. This is particularly relevant as agentic AI models become more advanced and conversational. Though they may appear human-like, savvy cardholders may find the interaction off-putting if they realize they are not talking to a human (or are not told up front), especially in situations that require reassurance, discretion, or accountability.
As issuers pursue cost efficiencies through automation and AI, these findings serve as a reminder that not every support interaction should be digital-first. Getting the balance right is key—not just for service satisfaction, but for long-term loyalty.
Survey Methodology
Mobile Pay Tracker
This Auriemma Group study was conducted online within the US by an independent field service provider on behalf of Auriemma Group (Auriemma) in April 2025 among 2,181 Mobile Pay (i.e., Apple Pay, Google Wallet, Samsung Wallet) eligible adult credit cardholders. The number of interviews completed for both is sufficient to allow for statistical significance testing among sub-groups at the 95% confidence level ±5%, unless otherwise noted. The purpose of the research was not disclosed, nor did respondents know the criteria for qualifying.
(London, UK): AI chatbots and Large Language Models (LLMs) such as ChatGPT have become commonplace in recent years to answer queries, whether that’s how to write a persuasive email or to request a replacement for a stolen card. According to Auriemma Group’s latest UK Cardbeat study, credit cardholders tend to be comfortable using AI chatbots to carry out basic tasks with their card issuer, but express skepticism when faced with more complex servicing needs.
For some time, AI Chatbots have been a key priority for card issuers to deliver a more streamlined customer service experience. However, through research and Roundtable discussions, Auriemma has found that the banks are shifting their focus towards AI development and innovation to support more efficient behind-the-scenes operations.
The complexity and difficulty of delivering a positive customer journey could be driving much of this shift. Auriemma Group’s latest UK Cardbeat study found that 65% of credit cardholders feel that AI chatbots often provide irrelevant or unhelpful responses. And what’s more, 57% said they regularly need to speak with a live representative following an AI-initiated customer service interaction.
Cardholders appear to be doubtful that chatbots can meet their needs, as only 36% say they wish their primary bank invested more heavily into chatbots, and just 26% feel chatbots are generally able to fully resolve issues.
While consumers believe AI chatbots still have much to improve, banks are identifying opportunities for AI to transform their business operationally. Members of Auriemma Group’s UK Roundtables recently shared they are leveraging or developing AI tools to create new efficiencies across Fraud, Disputes, Complaints, Customer Service and Collections Departments.
In fraud, current activities include using AI to score fraud cases and designing unique fraud warnings for customers. Firms are hoping AI can soon add randomness to fraud questioning when collecting personal customer data during the identification and verification (ID&V) process. A key initiative in development is writing a final response letter for customers who have submitted a complaint. Auriemma is also seeing firms use AI to transcribe customer service and complaints calls, as well as write summarizing notes in collections.
“Many of our members are changing their AI strategy to look more closely at back-office operations,” says Nicole Toussaint, Senior Manager at Auriemma Roundtables UK. “AI chatbots are very complex and require significant investment from a build and oversight perspective, but also come with the risk of providing misinformation. However, leveraging AI in other processes can simply make an employee or agent’s job more efficient without removing the human element from it entirely.”
Consumers will likely see chatbots become more sophisticated, but this may be just the tip of the iceberg compared to the unseen AI-led streamlined processes on the backend.
Survey Methodology
Cardbeat UK
This Auriemma Research study was conducted online within the UK by an independent field service provider on behalf of Auriemma in December 2023, among 800 adult credit cardholders. The number of interviews completed on a monthly basis is sufficient to allow for statistical significance testing between sub-groups at the 95% confidence level ± 5%, unless otherwise noted. The purpose of the research was not disclosed nor did the respondents know the criteria for qualification.