Social B(eye)as: Human and machine descriptions of people images
| Main Authors: | Barlas, PInar, Kyriakou, Kyriakos, Kleanthous, Styliani, Otterbacher, Jahna |
|---|---|
| Format: | Article |
| Terbitan: |
, 2019
|
| Subjects: | |
| Online Access: |
https://zenodo.org/record/3326827 |
Daftar Isi:
- Image analysis algorithms have become an indispens- able tool in our information ecosystem, facilitating new forms of visual communication and information shar- ing. At the same time, they enable large-scale socio- technical research which would otherwise be difficult to carry out. However, their outputs may exhibit social bias, especially when analyzing people images. Since most algorithms are proprietary and opaque, we pro- pose a method of auditing their outputs for social bi- ases. To be able to compare how algorithms interpret a controlled set of people images, we collected descrip- tions across six image tagging APIs. In order to com- pare these results to human behavior, we also collected descriptions on the same images from crowdworkers in two anglophone regions. While the APIs do not output explicitly offensive descriptions, as humans do, future work should consider if and how they reinforce social inequalities in implicit ways. Beyond computer vision auditing, the dataset of human- and machine-produced tags, and the typology of tags, can be used to explore a range of research questions related to both algorithmic and human behaviors.