Source: Antti Kauhanan and Petri Rouvinen, “Assessing Early Labor Market Effects of Generative AI, Evidence from Population Data,” (ETLA Working Papers, Helsinki, Finland, 2024).
Commentary: Using nationally representative data from the Incomes Register from Statistics Finland, authors Antti Kauhanen and Petri Rouvinen conducted a difference-in-difference analysis with data across all occupations, estimating the impact of generative AI exposure on wages for Finnish workers. The date of generative AI exposure was considered November 2022, the month ChatGPT was released. The authors used wage data from 2019 to 2022 to estimate wages before exposure to generative AI, and data from 2022 to 2024 to estimate wages after exposure. The researchers found that salaries increased by €78 per month on average across all occupations, the equivalent of 2 percent of the average salary in Finland. However, they also found that wages increased more in occupations where generative AI was most helpful in carrying out job tasks. Surprisingly, generative AI exposure had little to no effect on employment. These results could indicate that AI usage is a complement to human labor rather than a substitute, or that the impacts of generative AI on employment may not be realized for several years. It will be necessary to continue studying the impact of generative AI on wages and employment to fully understand its effects.
