How Do We Talk about Other People? Group (Un)Fairness in Natural Language Image Descriptions

By Jahna Otterbacher, Pinar Barlas, S. Kleanthous, K. Kyriakou. 2019

27 citations recorded by the source corpus.

Crowdsourcing plays a key role in developing algorithms for image recognition or captioning. Major datasets, such as MS COCO or Flickr30K, have been built by eliciting natural language descriptions of images from workers. Yet such elicitation tasks are susceptible to human biases, including stereotyping people depicted in images. Given the growing concerns surrounding discrimination in algorithms, as well as in the data used to train them, it is necessary to take a critical look at this practice. We conduct experiments at Figure Eight using a controlled set of people images. Men and women of various races are positioned in the same manner, wearing a grey t-shirt. We prompt workers for 10 descriptive labels, and consider them using the human-centric approach, which assumes reporting bias. We find that “what’s worth saying” about these uniform images often differs as a function of the gender and race of the depicted person, violating the notion of group fairness. Although this diversity in natural language people descriptions is expected and often beneficial, it could result in automated disparate impact if not managed properly.

Subfields: Natural Language Processing, Multimodal Learning, AI Ethics, Fairness & Society

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Natural Language Processing

In Multimodal Learning

In AI Ethics, Fairness & Society

Citations in the corpus

Understanding and Evaluating Racial Biases in Image Captioning (2021, background): “Prior works [54, 65] show that people are more likely to use racial descriptors when describing non-white individuals.”
Multimodal datasets: misogyny, pornography, and malignant stereotypes (2021, background): “…data dump of alt-text is besotted with high prevalence of issues such as missing important information, not being descriptive enough, resorting to stereotypical and offensive descriptors, being over descriptive (including ﬁlenames and special characters) or misrepresenting images [32, 30, 33].”
Taxonomizing and Measuring Representational Harms: A Look at Image Tagging (2023, methodology): “Another distribution-based approach compares the distributions of tags—either in a training dataset (Barlas et al. 2019a; Otterbacher et al. 2019; Wang et al. 2021; Yang et al. 2020) or output by an image tagging system (Alvi, Zisser-man, and Nellaker 2018; Kyriakou et al. 2019; Barlas et al.…”
Underspecification in Scene Description-to-Depiction Tasks (2022, background): “…is considered noteworthy or unexpected (Van Miltenburg et al., 2016; Misra et al., 2016), in social and cultural stereotyping (van Miltenburg, 2016; Zhao et al., 2017; Otterbacher et al., 2019), and in derogatory and offensive image associations (Birhane et al., 2021; Crawford and Paglen, 2019).”