![]() ![]() But in the ChatGPT era it also produces valuable data. In a world without AI chatbots, that would create what economists call productivity. “Let’s imagine you called me with a problem, and I solved it,” says Danielle Li, an economist at MIT’s Sloan School of Management who coauthored the study with MIT PhD candidate Lindsey Raymond and Erik Brynjolfsson, director Stanford’s Digital Economy Lab. But for the researchers conducting the study, the results raised a provocative new question: Should the top workers whose chats trained the bot be compensated? When the National Bureau for Economic Research, a nonprofit, published those results in late April, they were quickly seized upon as confirmation that ChatGPT-style bots would indeed transform work. ![]() ![]() And sure enough, when MIT and Stanford researchers analyzed the results, the AI tool had boosted the support team’s productivity by 14 percent. The bot had been trained on previous customer chats, with a special emphasis on answers from top performers. In 2020, 5,000 customer service agents mostly based in the Philippines became guinea pigs in an experiment testing a question that by 2023 would feel urgent: Can an AI assistant based on OpenAI’s text-generation technology make workers more productive? The automated helper offered agents suggested responses to small-business owners seeking tech support. ![]()
0 Comments
Leave a Reply. |