Prof. Li Lixing: AI’s Impact on Jobs
Nov 13-2023
On November 6, Prof. Li Lixing made a presentation on the impact of AI on employment at the 24th edition of NSD Academic Afternoon Tea, an event aiming to facilitate cross-disciplinary exchanges. He is NSD Professor of Economics and the research findings come from a research group, of which he is a member, on ChatGPT’s effects on jobs.
Prof. Li pointed out that digital technology has traversed the development stages of automation, informatization, and intelligentization. Its impact has stretched from manual to mental work, from simple to sophisticated work, and from factor reinforcement to process replacement. The emergence of large language model (LLM) AI, spearheaded by ChatGPT, heralds the extension of AI technology from specialized areas into generic ones, he said. Based on Skill Exposure Index, Prof. Li cited research which showed that 80% of the labor force will have at least 10% of their work affected by LLM during more than 50% of their working time, and 19% of labor force will have at least 50% of their tasks affected by ChatGPT during more than 50% of their working time. Scientific and critical thinking skills will be least susceptible to influences, while coding and writing skills will be most severely impacted. Compared with previous digital technologies, LLM will create a bigger shock to high-diploma, high-salaried groups.
Prof. Li also gave an in-depth introduction of the research group and their work. He said that the team sourced data from a large internet recruitment platform in the period between January 1, 2018 to April 19, 2023. All samples were randomly chosen based on yearly stratification, at around 200,000 pieces for each calendar year. After weighted adjustments, the total amount of valid samples came to approximately 1.02 million pieces, which cover 57 large professional categories and 1,027 job types. The team then built two LLM AI Skill Exposure Indices: detailed-job-activity based and task-based, and went on to analyze China’s job requirements and their connotational changes and compare the dynamic shifts in Chinese and American job markets.
Their research showed that over the last five years, newly created jobs in China have witnessed a decreasing skill exposure to LLM AI technologies, and an accelerated decline can be seen since 2022. The jobs that have borne the brunt are mainly white-collar ones, including accounting, translation, banking, sales, software and internet jobs. In 2021, companies showed an uptick in demand for LLM AI technologies, but it has gone down in the last two years, reflecting a delayed reaction to new technology development. Compared with other countries, China’s job market needs to strengthen its adaptability to new technologies, he said.
