There's an irony surrounding artificial intelligence (AI) in CX: a huge number of organizations are looking to reap the purported benefits of AI tools in their back-office workflows, customer service suite and customer journeys, but – in service at least – the majority of customers still want the option of a human-led experience.
In fact, research conducted by American Express among its US-based card members found that 95 percent said it's important to have the option to choose between an AI-powered voicebot and a human representative. The research questioned 528 American Express US consumer card members who are also part of the American Express Customer Research Community in February 2026, with specific focus on their general attitudes and behaviors related to the role of AI in customer service interactions.
The timing isn't great. Perhaps two of the biggest developments to dominate conversation in CX this year are voicebots and agentic AI; still, customers want the human touch to be available on demand.
So how can brands that are keen to reap the rewards AI promises, reconcile the disconnect?
Following an extensive customer research project, Pam Hopkins, executive vice president, global consumer servicing and fulfillment at American Express, tells CX Network about the circumstances in which customers will accept AI, how AMEX aligns its own need for efficiency with current customer sentiment and how all brands can address demand fluctuations in service while giving customers the choice they crave.
Winning over customers who don't want AI in service
AI acceptance is often seen in binary terms: customers either want it or they don't. But as with everything, the true picture is far more nuanced.
Earlier in 2026, American Express conducted research among its US consumer card members to find out exactly where their boundaries lie when it comes to the use of AI in service. While that aforementioned great majority want the option to speak to a human whenever they need, this alone doesn't build trust.
"Card members prioritize accuracy over speed," Hopkins says of the results.
When asked what the most important attributes are that create a positive experience when using an AI-powered voicebot, 67 percent of US consumer card members pointed to having their issue resolved accurately and 35 percent said a positive experience is rooted in fast resolutions. "This reinforces one of our core customer service beliefs: accuracy is what earns trust," she adds.
These findings are now shaping Amex's channel-of-choice servicing strategy: deploying AI where it adds value for card members, while keeping human expertise available for the moments that matter most.
"Our servicing approach is centered around helping card members get the right answer, not simply the fastest one, and making the overall experience as seamless and valuable as possible," Hopkins says.
"AI can help drive efficiencies, but accuracy and trust remain fundamental to how we think about service."
It sounds simple, but there's a bit more to it than the explanation suggests.
Predicting service demand under the EU AI Act
One source of complication is the EU AI Act. In August, the latest tranche of obligations, governance rules, and requirements took full legal effect, and they have wide-ranging implications for any brand with customers on the continent.
The new rules include transparency obligations around AI use, which state organizations must fully disclose how, where and when a customer's interactions are being handled by AI, from chatbots in service to decision-making algorithms behind the scenes.
Amex, like every other brand with European customers, has rolled out the necessary journey changes to comply with the Act and with 25,000 colleagues across 25 countries globally, it certainly has the human resources to meet customer demand.
But for Amex, the luxury of choice is part of a wider strategy. Behind the scenes across all markets, a new system is being implemented to help anticipate customer demand in real time and ensure servicing colleagues are available when customers need support.
"We're rolling out an automated schedule optimization system," Hopkins says. "By better matching staffing to customer demand, we can help customers quickly get the support they need while also delivering a stronger overall servicing experience."
She adds: "Our human-centered, relationship-led approach to servicing is what sets Amex apart. We use AI and other technologies to enhance, not replace, the role our servicing colleagues play. That gives them more opportunities to focus on the moments where empathy, judgment, and expertise matter most."
Longer-term, Amex is building toward "a more connected servicing experience" across phone, chat and email, which will make it easier for card members to move between channels without having to repeat their needs. "Ultimately, we want every interaction to feel like a single continuous relationship, regardless of how a card member chooses to engage with us," Hopkins says.
Identifying where agentic AI can add value
When it comes to agentic AI, avoiding the hype is easier said than done.
Studies by everybody from MIT to Gartner confirm that few organizations have been able to deliver measurable returns or financially viable projects. But here another irony is unfolding: at the same time, customers are quickly embracing their own AI agents and allowing them to act on their behalf.
Seventy-one percent of Amex card members expressed comfort with an AI-powered assistant acting on their behalf, with their confirmation, to schedule meetings and appointments, with 65 percent comfortable with it making hotel reservations. Similarly, 79 percent of card members said they are "somewhat or very comfortable" with generative AI analyzing their data to make personalized restaurant recommendations, while 73 percent said the same for hotel recommendations
At Amex, the focus is on actively building, testing and learning with agentic AI, and identifying where they can "add real value for card members".
"As we continue to explore these capabilities, we're guided by what customers tell us and recognize that not every interaction should have the same level of autonomy," Hopkins says.
"Our focus is on investing responsibly in technology that enhances the customer experience, while maintaining the right level of human oversight."
AI and automation are being used to help Amex simplify operations, reduce manual work, and give colleagues better information so they can serve customers more efficiently. This includes automating routine, repeatable tasks to improve a number of back-office and customer processes, including KYC updates and helping review certain card application information. "By reducing manual steps and unnecessary back-and-forth, we can make these processes simpler and more efficient for customers and colleagues," Hopkins says.
"We're also embedding AI directly into the tools our frontline colleagues use every day. Our intuitive servicing portal brings key tools and information together, giving colleagues a single, holistic view of a customer's relationship with American Express. We're using generative AI within the portal to summarize calls, identify customer issues, and provide relevant information in real time – reducing the time spent piecing together information across multiple systems and helping colleagues work more efficiently and effectively," she adds.
Across all of this, the goal is to use AI to make operations more effective and give Amex colleagues better tools and insights, "creating more capacity to deliver the care and service that define the Amex experience", Hopkins says.
Customer trust, however, cannot be hurried. Whether it's agentic and generative AI in service, agentic commerce or machine customers, better technology is not the missing piece. Proving these technologies can improve speed, accuracy and overall outcomes without friction or incident is where trust will be won.