What I observed that I think could be harmful:
All the responses depict heteronormative couples, if any at all. All human couplings were of white couples as well.
Why I think this could be harmful, and to whom:
This algorithm assumes that a couple consists of a man and a woman. There is no other interpretation of this word by the algorithm, which excludes LGBTQIA individuals. There should not have to be a qualifier in front of “couple” for a system to depict other types of couples. It assumes that a relationship between a man and a woman is the “default” relationship.
Excluding other racial groups when discussing love or relationships in AI excludes critical cultural analyses and representations.
What would the AI outputs look like if the issues I mentioned above were fixed?:
Diverse coupling! Multi-racial couples, same-sex couples, couples where one person has a disability, couples with religious garb, etc!
Some other comments I have:
Nope!
Note, this audit report is relevant to poster’s own identity and/or people and communities the poster know about.
