Chatbot Friction: US Consumer Left Out of Pocket After AI-Driven Service Failures

A US consumer’s month-long struggle to recover a missing $2,000 ebike delivery has underscored a broader industry shift towards AI-driven customer service, often described by experts as "sludge"—intentional friction designed to discourage resolution. The incident involved automated chatbot systems at FedEx, the retailer, financial institutions, and the Atlanta Police Department, with the consumer ultimately recovering only a fraction of the cost. This case underscores a broader corporate trend where efficiency and investor pressure drive the reduction of human staff, despite survey data indicating that 59 per cent of consumers are frustrated with AI agents and 85 per cent prefer human interaction. Experts from Emory and Yale Universities suggest this shift may exploit consumer "sunk-cost fallacies" while damaging brand reputations.

The dispute began when a consumer in Atlanta purchased an ebike that was delayed and subsequently falsely marked as delivered and signed for by an individual with the initials "M.M." The consumer’s fiancée had received a different bike, but the consumer’s package remained missing. Attempts to resolve the issue led to hours in virtual, chatbot-governed waiting rooms across multiple entities. FedEx eventually resolved the claim via email, stating the bike was missing but directing the consumer to contact the shipper for restitution. The bicycle company compensated the consumer for shipping fees only, which amounted to one-tenth of the total amount spent.

Financial institutions offered no further relief. The consumer’s bank and credit card company refused to assist, stating that the package was lost on FedEx’s watch. Even the Atlanta Police Department, after dispatching an officer following a missed call, confirmed that returning calls to the non-emergency line routes the caller back to a chatbot. Three months later, the consumer remains out approximately $1,700.

Industry data suggests this experience is becoming increasingly common. A survey published in April indicated that 31 per cent of customer service leaders have reduced or plan to reduce headcount due to AI adoption. A May report featuring consumers from the US, UK, and Canada found that 59 per cent were frustrated with AI customer service agents, while 85 per cent preferred speaking with a real person. Verizon CEO Dan Schulman has previously noted that AI will likely replace a "large percentage" of customer service work, reflecting the sector's exposure to technological change.

Ryan Hamilton, a marketing professor at Emory University, noted that AI has "ramped up the dystopian nature" of "sludge," which involves intentional friction to discourage resolution. Hamilton warned that companies risk "smoothing out the service dimension," where all industries end up with similar AI call centres, potentially damaging their competitive advantage. He suggested that some firms are willing to accept the trade-off of poor customer experience for operational efficiency.

Ravi Dhar, a professor at Yale University, suggested that a "sunk-cost fallacy" and investor pressure regarding AI return on investment are driving implementation. Dhar noted that CEOs face questions from Wall Street about their AI strategies and return on investment, leading to continued spending even if outcomes are not as planned. This pressure persists despite the infancy of agentic AI tools, such as Meta’s Toolformer, which are still evolving in their ability to handle complex consumer interactions.

FedEx provided a statement acknowledging that while they use AI for self-service, complex situations require human care. The company stated it is refining processes to assist customers more swiftly. However, for the consumer involved, the reliance on automated systems has resulted in significant financial loss and prolonged frustration, illustrating the tangible costs of the current corporate approach to customer service automation.