OBJECTIVE VS. SUBJECTIVE MEASURES OF OCCUPATIONAL AI EXPOSURE: A COMPARATIVE ANALYTICAL APPROACH
Maria URSU, Ludovic DIOSZEGI
Doctoral School of Economic Sciences, University of Oradea, Oradea, Romania maria.ursu@student.uoradea.ro
dioszegi.ludovic@student.uoradea.ro
Abstract: Perception of artificial intelligence (AI) represents an important indicator of how societies interpret and adapt to rapid technological change. While AI increasingly reshapes occupational structures, the divergence between objective measures of exposure and subjective perceptions remains insufficiently understood. Objective exposure can be estimated using occupational task profiles and quantitative indices, whereas subjective perception reflects deeper social, cultural and psychological mechanisms that influence how individuals internalize technological transformations. This study examines how young people evaluate AI exposure in relation to the occupations they aspire to, contrasting their assessments with established quantitative exposure indicators. Using survey data collected from 135 respondents aged 15–25, we compare perceived exposure with model-generated values to identify systematic gaps in understanding. Our results show that the quantitative model produces relatively consistent exposure levels across occupations, with most jobs falling within a moderate to high exposure range and displaying limited variation. By contrast, subjective perceptions are considerably more heterogeneous. Respondents generally assign lower exposure scores than the model predicts, indicating a tendency to underestimate the extent to which AI may affect future job content. Although the median perception score (6) is close to the model’s value, the mean is significantly lower (5.5 < 6.86), reflecting a distribution skewed toward lower perceived exposure. Gendered patterns also emerge: women report a higher median perception score, potentially reflecting greater concern regarding AI-driven disruption, while men express more optimistic assessments. These findings suggest that perceived vulnerability to AI is shaped not only by occupational characteristics but also by broader cultural and social factors. Overall, the study highlights a persistent misalignment between quantitative exposure indicators and human perception, raising important implications for career guidance, skills policy and public understanding of technological change.
Keywords: artificial-intelligence; perception; labour-market; employment; exposure.
JEL Classification: J24; O33; J21
