#44 Detecting Emergent Intersectional Biases: Contextualized Word Embeddings Contain a Distribution of Human-Like Biases
With the starting point that implicit human biases are reflected in the statistical regularities of language, it is possible to measure biases in English static word embeddings. State-of-the-art neural language models generate dynamic word embeddings dependent on the context in which the word appears. Current methods measure pre-defined social and intersectional biases that occur in contexts defined by sentence templates. Dispensing with templates, we introduce the Contextualized Embedding Association Test (CEAT), that can summarize the magnitude of overall bias in neural language models by incorporating a random-effects model. Experiments on social and intersectional biases show that CEAT finds evidence of all tested biases and provides comprehensive information on the variance of effect magnitudes of the same bias in different contexts. All the models trained on English corpora that we study contain biased representations. GPT-2 contains the smallest magnitude of overall bias followed by GPT, BERT, and then ELMo, negatively correlating with the contextualization levels of the models.Furthermore, we develop two methods, Intersectional Bias Detection (IBD) and Emergent Intersectional Bias Detection (EIBD), to automatically identify the intersectional biases and emergent intersectional biases from static word embeddings in addition to measuring them in contextualized word embeddings. We present the first algorithmic bias detection findings on how intersectional group members are strongly associated with unique emergent biases that do not overlap with the biases of their constituent minority identities. IBD achieves an accuracy of 81.6% and 82.7%, respectively, when detecting the intersectional biases of African American females and Mexican American females, where the random correct identification rates are 14.3% and 13.3%. EIBD reaches an accuracy of 84.7% and 65.3%, respectively, when detecting the emergent intersectional biases unique to African American females and Mexican American females, where the random correct identification rates are 9.2% and 6.1%. Our results indicate that intersectional biases associated with members of multiple minority groups, such as African American females and Mexican American females, have the highest magnitude across all neural language models.
#97 Fairness for Unobserved Characteristics: Insights from Technological Impacts on Queer Communities
Advances in algorithmic fairness have largely omitted sexual orientation and gender identity. We explore queer concerns in privacy, censorship, language, online safety, health, and employment to study the positive and negative effects of artificial intelligence on queer communities. These issues underscore the need for new directions in fairness research that take into account a multiplicity of considerations, from privacy preservation, context sensitivity and process fairness, to an awareness of sociotechnical impact and the increasingly important role of inclusive and participatory research processes. Most current approaches for algorithmic fairness assume that the target characteristics for fairness—frequently, race and legal gender—can be observed or recorded. Sexual orientation and gender identity are prototypical instances of unobserved characteristics, which are frequently missing, unknown or fundamentally unmeasurable. This paper highlights the importance of developing new approaches for algorithmic fairness that break away from the prevailing assumption of observed characteristics.
#227 Gender Bias and Under-Representation in Natural Language Processing across Human Languages
Natural Language Processing (NLP) systems are at the heart of many critical automated decision-making systems making crucial recommendations about our future world. However, these systems reflect a wide range of biases, from gender bias to a bias in which voices they represent. In this paper, a team including speakers of 9 languages – Chinese, Spanish, English, Arabic, German, French, Farsi, Urdu, and Wolof – reports and analyzes measurements of gender bias in the Wikipedia corpora for these 9 languages. In the process, we also document how our work exposes crucial gaps in the NLP-pipeline for many languages. Despite substantial investments in multilingual support, the modern NLP-pipeline still systematically and dramatically under-represents the majority of human voices in the NLP-guided decisions that are shaping our collective future. We develop extensions to profession-level and corpus-level gender bias metric calculations originally designed for English and apply them to 8 other languages, including languages like Spanish, Arabic, German, French and Urdu that have grammatically gendered nouns including different feminine, masculine and neuter profession words. We compare these gender bias measurements across the Wikipedia corpora in different languages as well as across some corpora of more traditional literature.