Most of us have been interacting with customer support agents for years. It can be a frustrating experience: Oftentimes the agent knows less than we do about their product or service, calls are dropped or transferred to other agents. About two years ago, I had such a bad experience with AT&T Inc.’s customer support that I ended up cancelling my cell and internet service with the company.

But now there are artificial intelligence chatbots and chat programs that are supposed to make our lives better. With all the attention focused on ChatGPT and other AI-based chatbots, a new long-term research study has found that AI can help improve support, but only under carefully controlled situations. Let’s examine the specific circumstances and what’s in store for the future of support.

The paper was sponsored by the National Bureau of Economic Research and written by Erik Brynjolfsson of Stanford University’s Digital Economy Laboratory and Lindsey Raymond and Danielle Li of MIT’s Sloan School of Management. They examined for several months more than 5,000 mostly Philippine-based customer support agents working at a Fortune 500 enterprise business process software provider. The agents made use of an AI tool to help guide their support conversations with customers, meaning that the agents are free to use or ignore the AI’s suggestions.

They found a 14% increase in employee productivity in answering customer support  queries, particularly among the less skilled and newly hired workers. These folks learned how to operate at higher levels by using the AI tools. Interestingly, more experienced workers had less to gain from the AI tools, mainly because of how the AI captures what these workers already know and use during their calls. “While lower-skilled workers improve from having access to AI recommendations, they may distract the highest-skilled workers, who are already doing their jobs effectively,” the authors said in their paper.

Perhaps the most interesting finding is how AI can help improve how customers treat the support agents, resulting in fewer requests for frustrated customers to escalate their issue by asking to speak to a supervisor. But the survey is most notable in how it dissects the way people work and how AI can coexist and help improve productivity.

We have written plenty on the effect of AI on the workplace, including most recently this post about best AI practices and this post from 2019 about how IBM is upskilling employees with AI. That post didn’t age well: IBM’s latest plan is to lay off 7,800 workers and replace them with AI.

How do these results influence the best way to deploy AI-based chat tools? Here are a few of my own recommendations:

1. The best productivity gains happen when humans and AI can work together. Deploying AI is not an either/or situation. Learning how to bridge both sides will be an important deployment skill in our future.
2. Part of these gains involve reducing employee turnover and training costs, which gets to understanding the causes of these turnovers. Working in a call center is not a job for everyone. It can be high-stress and require time management skills. Having an expert AI chatbot to guide interactions can make the job more satisfying and less stressful, which helps to retain staff.
3. The best applications of AI-based tools is with repetitive tasks that involve matching patterns or following a sequence of actions. Workplaces that have a relatively stable product and predictable support issues work best at using AI.
4. Jobs that involve a combination of detailed product knowledge, problem-solving skills, and the ability to deal with frustrated customers are ripe for AI-based tools that can incorporate these elements.

Granted, the survey is relevant for a particularly narrow slice of the corporate workforce. But there are some solid lessons on how to incorporate AI into a business, and it’s a useful starting place for how to evaluate future AI efforts.

Image: TheDigitalArtist/Pixabay