What is it about?
This article argues that current AI ethics frameworks are too procedural, focusing heavily on checklists, technical metrics, and institutional compliance while ignoring how humans emotionally experience technology. It highlights a major asymmetry: massive resources are poured into designing AI that can detect or mimic human emotion, yet dominant governance rules offer no category to evaluate affective harm.
Featured Image
Photo by Gabriele Malaspina on Unsplash
Why is it important?
This paper exposes critical blind spots in how the tech industry currently reviews and audits AI systems, it exposes the limits of strict compliance, identifies affective harm and deconstruction of flawed tech assumptions.
Perspectives
Tech teams must look past basic data diversity and actively question whose expressive baseline they are building into code, ensuring they do not mistake a specific cultural norm for universal neutrality. Rather than evaluating AI solely in a vacuum, system defenders need to include experience-centered evaluations to track the psychological impact on vulnerable populations, while maintaining empathetic "human-in-the-loop" backups. Academics are urged to move beyond listing high-level moral principles and start treating human affect and culture as core analytical categories essential for true accountability.
Sharon Tettegah
University of California Santa Barbara
Read the Original
This page is a summary of: Lived, affective, and cultural dimensions of ethical AI: beyond procedural AI ethics, AI & Society, June 2026, Springer Science + Business Media,
DOI: 10.1007/s00146-026-03165-9.
You can read the full text:
Contributors
The following have contributed to this page
