- 25,000 workers only saw a 2.8% time saving after using GenAI chatbots
- 8.4% saw new tasks being created, adding to their workloads
- Time savings rarely translate to higher earnings
According to a new working paper from the University of Chicago’s Becker Friedman Institute for Economics, generative AI chatbots like ChatGPT, Gemini, and Claude have had minimal to no impact on wages or employment so far, despite initiation concerns and worker apprehension.
The findings come from a study of around 25,000 workers in Denmark across 11 AI-exposed occupations. The study found no significant changes in earnings or hours worked after the implementation of these AI tools.
Anders Humlum and Emilie Vestergaard, the paper’s authors, also noted no noticeable economic outcomes, such as total employment and wage bills at the firm level.
GenAI isn’t actually affecting workers that much, study finds
The paper explores how companies can guide their workers, noting that encouragement and training efforts can boost adoption. Firm-led investments nearly doubled adoption rates from 47% to 83%, according to the research.
However, the average time savings reported by users was only 2.8% – just over an hour on the basis that an employee is working a 40-hour week. Furthermore, only 8.4% of workers saw new jobs being created, such as teachers monitoring AI-assisted cheating, workers editing AI outputs and crafting better prompts.
Contrary to the time-saving promises, Humlum and Vestergaard noted that these additional responsibilities actually increased workloads in some cases, meaning that time savings only translated into higher earnings 3-7% of the time.
The researchers noted: “While AI chatbots are now widely used – saving users time and creating new job tasks, especially in workplaces that encourage their use – their overall impact on the labor market remains limited.”
They also noted that “labor market rigidities appear to delay the economic impact,” demonstrating how a company’s approach to artificial intelligence can dictate its success.
