More companies replace customer service staff with AI but often fail to improve service quality

More Danish companies are using AI chatbots in customer service, yet poor implementation risks damaging customer experience, researchers warn.

A new analysis by Dansk Erhverv shows that 15 percent of Danish businesses now use chatbots for customer service, up from six percent in 2023. But while AI can handle inquiries 24/7 without overtime costs, many companies have rushed adoption at the expense of service quality, according to researchers at Copenhagen Business School (CBS).

“They’ve taken the wrong approach,” said Torsten Ringberg, a CBS professor specializing in artificial intelligence. “They’re cutting costs instead of improving the customer experience.”

Christian Hendriksen, a CBS lecturer researching AI implementation, agreed, calling the quality of many language models “surprisingly poor.” He attributed this to companies opting for cheap, underdeveloped solutions.

“There’s no plug-and-play fix,” Hendriksen said. “Businesses need to invest time and resources to train AI properly—otherwise, it fails in real customer interactions.”

AI stumbles in high-profile cases

Several companies have publicly struggled with AI-driven customer service. Swedish fintech Klarna announced in early 2024 that its AI could replace 700 employees, handling 75 percent of customer chats. But within a year, the company reversed course, rehiring staff after admitting the technology fell short.

“In a world of AI, nothing will be as valuable as humans,” Klarna CEO Sebastian Siemiatkowski wrote on X (formerly Twitter). In a May 2025 interview with Bloomberg, he emphasized the importance of reassuring customers that human support remains available: “For our brand, it’s critical to signal that a person is always there if needed.”

Danish payment startup Dineros also claimed its AI handles over 71 percent of customer inquiries, saving the equivalent of 30 full-time positions. Founder Martin Thorborg argued the shift wasn’t about cost-cutting but efficiency, citing 24/7 availability.

Yet customer reviews on platforms like Trustpilot reveal widespread frustration with bot interactions, suggesting many companies have misjudged the balance between automation and service quality.

Some success—but no shortcuts

Not all AI deployments have faltered. Banks like Saxo and Nordea use chatbots for basic queries, while others remain cautious. Kasper Lynge Jacobsen, AI chief analyst at Dansk Erhverv, acknowledged mixed results but noted well-implemented systems can work—if companies commit to proper training and data investment.

“Everyone’s figuring it out as they go,” Jacobsen said. “The key is recognizing that AI isn’t a quick fix—it demands real effort to get right.”