Algorithmic bias? An empirical study into apparent gender-based discrimination in the display of STEM career ads

By Anja Lambrecht, Catherine E. Tucker. 2019

119 citations recorded by the source corpus.

We explore data from a field test of how an algorithm delivered ads promoting job opportunities in the Science, Technology, Engineering and Math (STEM) fields. This ad was explicitly intended to be gender-neutral in its delivery. Empirically, however, fewer women saw the ad than men. This happened because younger women are a prized demographic and are more expensive to show ads to. An algorithm which simply optimizes cost-effectiveness in ad delivery will deliver ads that were intended to be gender-neutral in an apparently discriminatory way, due to crowding out. We show that this empirical regularity extends to other major digital platforms.

Subfields: Economics and Econometrics, Marketing, Sociology and Political Science, Safety, Robustness & Interpretability

How researchers see this paper

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

In Economics and Econometrics

In Marketing

In Sociology and Political Science

In Safety, Robustness & Interpretability

Citations in the corpus

Reducing Gender Bias in Word-Level Language Models with a Gender-Equalizing Loss Function (2019, methodology): “We apply CDA where we swap all the gendered words using a bidirectional dictionary of gender pairs described by Lu et al. (2018). This creates a dataset twice the size of the original data, with exactly the same contextual distributions for both genders and we use it to train the language models.”
How We've Taught Algorithms to See Identity: Constructing Race and Gender in Image Databases for Facial Analysis (2020, background): “Providing documentation on these decision-making processes would make databases more transparent, and thus more usable to third-parties.”
The adoption of artificial intelligence in employee recruitment: The influence of contextual factors (2021, background): “Several studies have illustrated AI’s implications in HR recruitment, with focuses on candidates’ attitudes toward AI (Suen et al., 2019; Van Esch et al., 2019) and the fairness of AI recruitment (Lambrecht & Tucker, 2019; Suen et al., 2019).”
Artificial Intelligence and Marketing: Pitfalls and Opportunities (2020, background): “For instance, in online targeted advertising, Lambrecht and Tucker (2019) demonstrated that since women are more highly priced target consumers than men, real-time-bidding competition is fiercer.”
Mitigating Gender Bias in Captioning Systems (2020, background): “Gender Bias in Dataset: The issue of gender bias has been studied in a variety of AI domains [3, 5–8, 12], especially in natural language processing domain [15, 19, 32, 35, 37, 44, 45].”
Adoption and use of AI tools: a research agenda grounded in UTAUT (2021, background): “…of such AI-based tools, especially in comparison to previous approaches and algorithms, is of interest (Razzaghi et al. 2019), with a particular focus on avoiding biases that can creep into models, especially when learning comes from data that is riddled with bias (Lambrecht and Tucker 2019).”
A strategic framework for artificial intelligence in marketing (2020, background): “Researchers have also shown that gender bias can occur without any conscious (or unconscious) attempt to produce a biased outcome — using only an unbiased algorithm (Lambrecht and Tucker 2019).”
How Computers See Gender (2019, background): “However, the caveat to ignoring gender in datasets is potentially reifying gender bias (e.g. ([24, 60])).”
Individual Fairness in Pipelines (2020, background): “RELATED WORK There is a wide variety of work concerning fairness in machine learning [9, 14, 25, 17, 3, 18, 26, 22, 10, 16, 15, 11, 12, 20, 19, 4, 24, 5].”
Bias and Debias in Recommender System: A Survey and Future Directions (2020, unknown): “imination and reduced the visibility for disadvantaged groups (e.g., under-representing the minorities, racial or gender stereotypes). For example, in the context of job recommendation, previous work [52], [53] found that, compared to men, women saw less ads about high paying jobs and career coachin”
Algorithmic bias: review, synthesis, and future research directions (2021, background): “…research shows that algorithm-delivered ads promoting job opportunities in the science, technology, engineering, and mathematics fields tend to be displayed to men more frequently than to women simply because the algorithm’s objective is to optimise the advertising costs (Lambrecht & Tucker, 2019).”
The ethics of algorithms: key problems and solutions (2020, unknown): “Bias has also been reported in algorithmic advertisement, with opportunities for higherpaying jobs and jobs within the field of science and technology advertised to men more often than to women (Datta et al. 2015; Lambrecht and Tucker 2019).”