Bias in AI-generated Images

AI-generated images display patterns of sexist, racist, and class-based prejudices. Eva Gengler has published a study on this.

Her overview (on LinkedIn) of the results:

Gender Bias:

Men are overrepresented in contexts of power and success.
Women are mainly depicted in beauty-related contexts, often sexualized.
In the context of power, older men and younger women are most commonly portrayed.

Racial Bias:

People of color are underrepresented, particularly in positions of power and success.
White people dominate in all categories.
A slight increase in diversity is seen in beauty contexts.

Class-based Bias:

The majority of depicted individuals come from privileged socioeconomic backgrounds with a wealthy appearance (business attire, elegance).
There is low representation of working-class people or “average” individuals (no model career, etc.).

Eva Gengler presented on *”Feminist AI for a Fairer World”* at the herCAREER Expo 2024. “We have the great opportunity today to develop and use systems, which are being deployed everywhere, with the goal of changing power.”
She was also one of the speakers at the discussion event *”Our Future of Work with Artificial Intelligence – How Women Can Benefit,”* presented by the Bavarian State Ministry for Family, Work, and Social Affairs. The discussion will soon be available as a podcast (link in the comments)!
Additionally, Eva served as a Table Captain at herCAREER@Night

Thank you, dear Eva, for your active role in the herCAREER community on such exciting topics!