As more companies adopt Generative AI (GenAI), the potential to significantly alter the job market continues to increase. Goldman Sachs forecasts that up to 300 million jobs could be lost or downgraded due to this fast-growing technology. McKinsey predicts that GenAI will force 12 million job switches and automate 30% of hours worked in the US by 2030. In May, AI was responsible for nearly 5% of job cuts in the US.
A recent study by Paweł Gmyrek, Janine Berg, and David Bescond, published in the International Labour Organization Working Paper, offers a more optimistic perspective about the impact of GenAI on jobs, contrary to the prevalent doom-and-gloom predictions and actual job losses. The study utilized the GPT-4 model to calculate potential exposure scores at the task level, and then projected likely employment impacts at both the global scale and by country income group.
According to the findings, clerical work is the occupation most vulnerable to GenAI, with 24% of clerical tasks being highly exposed and an additional 58% having medium-level exposure. These findings support earlier reports on the impact of AI on clerical roles. However, for other occupational groups, the exposure to GenAI is lower. The highest proportion of highly exposed professions ranges between 1% and 4%, and tasks with medium exposure do not exceed 25%.
The most notable effect of GenAI is expected to be the improvement of work by automating specific tasks within a job, thereby allowing more time for other duties, rather than fully automating entire jobs. The potential employment effects vary greatly across country income groups due to differences in occupational structures. In low-income countries, only 0.4% of total employment is potentially exposed to automation effects, while in high-income countries this figure rises to 5.5%.
The greater impact comes from augmentation, which can affect 10.4% of employment in low-income countries and 13.4% of employment in high-income countries. However, these effects do not consider infrastructure constraints, which may hinder the use of GenAI in lower-income countries and potentially widen the productivity gap.
Although the study offers some hope, the authors emphasize that the main value of their analysis lies not in the exact estimates…