Generative AI Can Make Some Pros Much More Productive, Study Finds


When thinking about generative AI in the workforce, it’s easy to imagine the worst-case scenario: AI replacing human jobs. However, a study shows that generative AI tools can have a positive impact on professionals, especially those working in the customer service industry.

A working paper from the National Bureau of Economic Research (USA) shows that access to generative AI can increase employee productivity by an average of 14%. The unit of measure is the number of customer issues agents were able to resolve per hour.

To conduct this study, the NBER used data from 5,000 customer support agents working for a Fortune 500 software company. The agents used a tool based on a recent version of the Great Language Model (LLM ) Generative Pre-trained Transformer (GPT) from Open AI to help them in their work.

Productivity increases have disproportionately benefited less skilled and less experienced workers

The LLM monitored customer chats in real time to provide agents with suggestions on how to respond. This allowed agents to respond faster, respond to more chats per hour, and resolve chats better, according to the study.

However, the study found that the increase in productivity disproportionately benefited less skilled and less experienced workers. AI tools have indeed made it possible to fill the lack of experience. Workers using AI with two months tenure performed just as well as professionals with six months tenure and not using AI.

Therefore, the study showed that highly skilled professionals do not have as much to gain from using AI assistance, since AI recommendations essentially mimic knowledge that professionals highly qualified already possess.

Reduce reliance on supervisor

Besides maximizing the productivity of pros, AI can also help improve the way they are handled, reducing the likelihood of calls being routed to a supervisor.

Overall, this study provides a concrete example of how AI can be used to help some pros instead of just replacing them.


Source: “ZDNet.com”



Source link -97