Redesigning AI for Inclusion

Abstract

The current study examined how AI systems marginalize students from diverse cultural and linguistic backgrounds and proposed strategies to enhance inclusivity in AI-driven educational environments. Using a phenomenological design, in-depth semi-structured interviews were conducted with 11 university students. The data were analyzed using a thematic data analysis technique. The findings showed that students liked how AI systems made learning more personalized. However, most of them reported cultural biases in how the AI systems gave feedback and predicted their performance. The students reported that automated feedback often failed to appreciate context-specific linguistic variations, labeling non-standard expressions or culturally differentiated arguments as “poor grammar” or “unclear thinking.” Content suggestions in learning platforms also appeared to support Western-centric narratives. Several participants reported that AI-generated predictions regarding their academic performance did not match their actual learning efforts. This chapter had several practical implications.

Ähnliche Arbeiten