A new study has revealed that artificial intelligence cannot work efficiently enough in call centers. A new study conducted in China has revealed that AI-powered assistants used in call centers, far from increasing efficiency, increase the error rate and stress levels of employees. A field study conducted by Chinese researchers with an energy company has revealed in detail the problems experienced by customer representatives with these systems.

Artificial intelligence cannot work efficiently in call centers

Artificial intelligence systems experience serious problems, especially in the process of converting speech to text. The study stated that elements such as customers’ accents, fast speech tempos or numerical expressions are among the points where the system has the most difficulty.

Similarly, homophones and nuances in the natural flow of language are also among the areas that AI systems cannot understand. This situation increases the workload of representatives and causes them to try to fix the system.

The systems’ emotion recognition capabilities are also considered unreliable by employees. Customer representatives interviewed as part of the research stated that AI labels ordinary conversations with negative emotions such as anger or dissatisfaction, and that this is usually due to misinterpretation of tone of voice. Employees state that instead of relying on these analyses, they prefer to understand the customer’s mood through their own experiences and intuition.

Although AI systems seem to save time in some routine writing tasks, semantic errors, unnecessary repetitions and missing information in the content produced lead agents to re-edit these contents. This causes the system to create extra work instead of contributing.

The report emphasizes that technological integration creates not only technical but also psychological and sociological problems. There is a visible resistance among employees to the use of AI. Fear of losing their jobs, dependence on AI analyses in performance evaluations and the distrust created by the system increase the general stress level of call center employees.

There is a similar dissatisfaction on the customer side. The research shows that many customers do not accept a customer service structure that is overly dependent on AI. A more recent study states that 40 percent of such projects are expected to be canceled in the next few years.