Yong Suk Lee, an assistant professor in the Keough School of Global Affairs, studies AI ethics, competition and nationalism, regulatory issues and the implications for labor and organizations. He considers the rapid adoption of large language models (LLMs) like ChatGPT as one of the key developments of 2023 — one that can both complement the labor market by performing specialized tasks or stunt it by replacing human workers.
Lee says OpenAI may have jumped the gun in launching ChatGPT, but many companies and organizations are now racing to build their own custom LLMs that can serve as alternatives to that platform or that can perform in specialized areas such as medicine, coding and even math problem solving.
“Much of the attention and research have been focused on the development of these new LLMs, but relatively less research has been done on how we should use LLMs, the societal consequences of these models and potential policy recommendations,” Lee explained.
One major concern Lee sees is whether LLMs will replace human workers. “Virtually all writing tasks can be complemented by LLMs to some degree,” Lee said. “People are already using ChatGPT to assist in writing emails and speeches, planning itineraries and writing business plans, etc. As of now, LLMs mostly assist workers in specific tasks and are unlikely to fully replace a worker. However, as the economy slows, firms may seek ways to increase productivity and cut costs by using LLMs. Rather than hiring additional workers, firms could augment existing workers with AI and LLMs, ultimately reducing labor demand.”
Lee concluded that we will have to wait and see whether 2023 will also be remembered as the year firms started to slow hiring because of LLMs. “For now, to be competitive in the labor market, workers may need to be proficient in using and prompting LLMs and, at the same time, be familiar with their limitations,” he said.