Usage of a generative AI assistant tool in call centers resulted in an average productivity boost of 13.8%, measured by the number of resolved customer issues per hour, according to a recent study conducted by researchers from Stanford Digital Economy Laboratory and MIT Sloan School of Management.
The research sheds light on the impact of generative AI in workplace environments, specifically within the customer service sector, which already boasts high adoption rates of AI technology.
The study revealed that the productivity improvements were more pronounced in less skilled and less experienced agents. The AI assistant was observed to help these workers improve at a faster pace, enabling agents with two months of experience to perform as effectively as those with six months of experience who did not use the AI assistant.
The AI system’s ability to share knowledge across workers could be attributed to its capability to capture tacit knowledge and provide real-time recommendations more efficiently than a manager.
Experienced workers benefit less from AI tools
Interestingly, researchers found that higher skilled and more experienced workers showed less improvement in productivity.
They theorized that this could be because the AI recommendations, which are trained using data from experienced workers, are already inherent in the skills and wisdom of these higher-skilled workers. This raises questions about whether high-skilled workers should be compensated for the quality training data they generate for AI systems.
Additionally, the study noted improvements in the way customers treated agents who learned the job faster with the aid of the AI assistant. This highlights the potential of generative AI in fostering positive interactions between customers and call center agents.
More research needed
The researchers acknowledged the limitations of their study, stating that their findings do not capture potential long-term impacts on skill demand, job design, wages, or customer demand.
They suggested that more effective technical support could lead to contact center agents taking on more complex responsibilities, potentially increasing overall demand even if agents become more productive.
Furthermore, the researchers emphasized that the study’s insights are specific to an industry with a relatively stable product and set of technical support questions. In rapidly changing environments, the value of AI recommendations, which are trained on historical data, may differ significantly.