Dive Brief:
- Customer support agents who were given access to a generative artificial intelligence assistant increased their productivity by 14% on average, according to a group of researchers at Stanford University’s Digital Economic Laboratory and the Massachusetts Institute of Technology.
- The researchers, whose findings were published in a National Bureau of Economic Research working paper, studied the deployment of an AI-based chat assistant built on a large language model developed by OpenAI. The study encompassed more than 5,000 agents working for a Fortune 500 firm, and the chat assistant monitored customer service chats while providing agents with real-time response suggestions.
- Overall, the assistant decreased the time it took agents to handle individual chats, increased the number of chats agents were able to handle per hour and slightly increased the number of successfully resolved chats. However, the researchers said AI use “disproportionately” increased performance for less skilled and less experienced agents, with “minimal impact” on experienced and highly skilled workers.
Dive Insight:
It took less than six months for generative AI tools like ChatGPT and Midjourney to gain widespread public access and use. Now, employers may be seeing the very first documented use cases for such tools in the workplace.
The NBER study’s authors noted that functions like customer service have been popular use cases for AI in the past. The study cited a 2021 McKinsey survey that found 27% of respondents whose organizations had adopted AI used it for service operations optimization, and 22% said the same of contact center automation.
Though the NBER study found that the chat assistant helped newer customer support agents perform at levels comparable to more senior agents, the researchers said they found few positive effects of the tech on higher-skilled and more experienced agents.
“We posit that high-skill workers may have less to gain from AI assistance precisely because AI recommendations capture the potentially tacit knowledge embodied in their own behaviors,” the researchers wrote. “Rather, low-skill workers are more likely to improve by incorporating these behaviors by adhering to AI suggestions.”
As mainstream technology firms move to incorporate generative AI into products that workers already use, it may be important for employers to have an understanding of the extent to which employees may already be using AI. A recent Fishbowl survey found that, of workers using AI tools for work-related tasks, 68% said they did not inform their bosses before doing so.
And when deciding how and when to implement emerging AI tech, worker needs should drive the conversation, consulting firm Bain & Co. said in a March report.
