A new study that the Danish labor market analyzes in 2023 and 2024 suggests that generative AI models such as chatgpt have had almost no significant impact on total wages or employment, despite rapid acceptance in some workplaces. The findings, detailed in a working document from economists from the University of Chicago and the University of Copenhagen, offer an early, large -scale empirical view of the transforming potential of AI.
In “Great Language Models, Small Labor Market Effects”, Es Anders Humlum and Emilie Vestergaard focused specifically on the impact of AI Chatbots on 11 professions that are often considered vulnerable to automation, including accountants, software developers and specialists in the field of customer support. Their analysis included data from 25,000 employees and 7,000 workplaces in Denmark.
Despite finding widespread and often accepting these tools by the employer, the study concluded that “AI-Chatbots had no significant influence on the income or hours included in a profession” during the period studied. The confidence intervals in their statistical analysis made average effects greater than 1 percent.
“The approval of these chatbots has been remarkably fast,” Humlum told the register on the study. “Most employees in the exposed professions have now taken over these chatbots … But when we look at the economic results, it really did not move the needle.”
Create more work?
During the study, the researchers investigated how business investments in AI influenced the acceptance of employees and how chatbots changed workplace processes. While business investments increased the acceptance of AI tools – saving time for 64 to 90 percent of users in professions studied – the actual benefits were less substantial than expected.
The research showed that AI chatbots actually made new task tasks for 8.4 percent of employees, including some who have not used the tools themselves, which compensates for potential time savings. For example, many teachers now spend time detecting whether students use chatgpt for homework, while other employees assess AI output quality or try to make effective instructions.
