AI models fall short in job market predictions for unemployed individuals
A recent study utilising Finland’s population data reveals that AI models perform inconsistently in predicting employment, and are unable to capture the life course patterns for the people who end up unemployed.
"This deficiency of AI models means that government organisations should be careful when using AI in social policy making", says Sanni Kuikka, PhD student in sociological demography, who conducted the study.
The study finds that AI models perform inconsistently across various groups. They are accurate in predicting the employment status of the already employed while incorrectly predicting most of the unemployed.
Additionally, the predictions varied significantly based on socioeconomic background. The outcomes for individuals with low parental education were overall more difficult to predict than the outcomes for individuals with mid or high parental education.
More about the study
- The study The (Un)Predictability of Early (Un)Employment: A Machine Learning Approach is published in Socius: Sociological Research for a Dynamic World.
- Sanni Kuikka's profiile page
Last updated: November 14, 2024
Source: The Department of Sociology