- A highly skilled customer service worker found that AI assistants weren’t really conducive to productivity.
- Conducted at an unnamed Fortune 500 company, the study was conducted by researchers at Stanford University and MIT.
- The study found that AI can help less-skilled workers “communicate like highly-skilled agents.”
New research may provide insight into what generative AI means for workers of different skill levels.
In a recent case study of more than 5,000 customer service representatives from Fortune 500 companies, highly skilled agents were trained in an AI chat program aimed at helping them provide better, faster answers to customers. did little to improve productivity.
Erik Brynjolfsson, a professor at Stanford University’s Human-Centric AI Institute and one of the study’s authors, told Insider that the study found that productivity among highly skilled workers was “nearly 0%.” says it has improved.
Chat programs deployed by software companies were intended to provide customer representatives with “real-time suggestions” for handling customer interactions. It delivered an average productivity gain of 14%, but the study found that the gain was felt most by lower-skilled workers.
Brynjolfsson told Insider, “The system is really taking a lot of the expertise, the tacit knowledge of highly skilled workers, and making it available to less experienced workers.” I’m here. “But as you might imagine, it’s not as useful for highly skilled workers, because that’s what they already know.”
Researchers say AI tools are trained to mimic the effective customer conversation styles exhibited by highly skilled employees. Research shows that the AI tool in question was trained to understand features such as call duration and call outcome, and how top performers behaved on these aspects.
“We found that AI models were trained on human-generated data in environments with high variability in individual agent performance,” the study authors wrote. “As a result, when the model identifies patterns that distinguish between successful and unsuccessful calls, it is implicitly learning the differences that characterize high-skilled and low-skilled workers.”
“Our overall findings show that generative AI working alongside humans can have a significant positive impact on the productivity and retention of individual workers,” said the study. The author writes
While the results show potential benefits for less-skilled customer support agents from using AI assistance, Brynjolfsson told Insider, “High-skilled employees will also benefit. I can imagine how,” he said.
The study notes that customer service is a particularly mature area where companies are deploying artificial intelligence technology.
“From an AI perspective, a customer-agent conversation can be thought of as a series of pattern-matching problems to find the optimal course of action,” the study’s authors noted, adding that AI tools like It evokes how it reflects the method. ChatGPT works by predicting text.
The AI in this case involved a system focused on customer conversations that incorporated “the latest version of GPT with additional ML algorithms,” citing machine learning algorithms, the report said. says.
“I think we are all starting to realize that generative AI is different than some of the previous waves,” said Brynjolfsson.
He said there are still problems with “hallucinations and facts”, referring to the phenomenon of AI tools authoritatively providing false responses. I added that it may serve the purpose.
Although there are still issues such as incorrect answers, he thinks “having a human in the loop” is a good thing. This may give some people who are worried about what AI means for job security, at least some peace of mind for now.
“A human-machine combination is better than a machine alone or a human alone,” said Brynjolfsson.
Watch Now: Top Insider Inc. Videos
Now loading…
