Stereotypes, disproportions, and power asymmetries in the visual portrayal of migrants in ten countries: an interdisciplinary AI-based approach

This study uses an interdisciplinary AI-based approach to analyze the visual portrayal of migrants in media from ten countries. Deep learning techniques quantify demographic and emotional information in images associated with migrants, refugees, and expats. Results reveal that portrayals predominantly focus on asylum seekers, associating them with poverty and risks for host societies. Demographic portrayals mismatch official statistics, with expats over-representing "white" faces and under-representing "Asian" faces, while migrants and refugees align with low-skilled migrant demographics. The study also exposes the power struggle inherent in the "expat vs. migrant" dichotomy and its colonial nature. Emotions displayed are predominantly negative, aligning with stereotypes, while positive emotions are associated with women, and expats more than refugees and migrants. The study confirms previous findings on under-representation of migrant women and highlights differences in portrayal across locations.

Juan Sebastian Olier, Camilla Spadavecchia