Automated Experiments on Ad Privacy Settings
By Amit Datta, Michael Carl Tschantz, Anupam Datta. 2014
784 citations recorded by the source corpus.
Abstract To partly address people’s concerns over web tracking, Google has created the Ad Settings webpage to provide information about and some choice over the profiles Google creates on users. We present AdFisher, an automated tool that explores how user behaviors, Google’s ads, and Ad Settings interact. AdFisher can run browser-based experiments and analyze data using machine learning and significance tests. Our tool uses a rigorous experimental design and statistical analysis to ensure the statistical soundness of our results. We use AdFisher to find that the Ad Settings was opaque about some features of a user’s profile, that it does provide some choice on ads, and that these choices can lead to seemingly discriminatory ads. In particular, we found that visiting webpages associated with substance abuse changed the ads shown but not the settings page. We also found that setting the gender to female resulted in getting fewer instances of an ad related to high paying jobs than setting it to male. We cannot determine who caused these findings due to our limited visibility into the ad ecosystem, which includes Google, advertisers, websites, and users. Nevertheless, these results can form the starting point for deeper investigations by either the companies themselves or by regulatory bodies.
Subfields: Clinical Psychology, Marketing, Sociology and Political Science, Privacy & Security, 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 Clinical Psychology
- Catherine E. Tucker
- Eirini Ntoutsi
- Wenbin Zhang
- Bo Cowgill
- L. Elisa Celis
- Nisheeth K. Vishnoi
- Anupam Datta
- Saif M. Mohammad
In Marketing
- Catherine E. Tucker
- Eirini Ntoutsi
- L. Elisa Celis
- Nisheeth K. Vishnoi
- Bo Cowgill
- Wenbin Zhang
- Alan Mislove
- Anja Lambrecht
In Sociology and Political Science
- Aleksandra Korolova
- Alexandra Olteanu
- Carlos Castillo
- Fernando Diaz
- Catherine E. Tucker
- Nisheeth K. Vishnoi
- L. Elisa Celis
- Anupam Datta
In Privacy & Security
- Aleksandra Korolova
- Cynthia Dwork
- Anupam Datta
- Catherine E. Tucker
- Boi Faltings
- Eirini Ntoutsi
- Jonathan Ullman
- Kamalika Chaudhuri
In AI Ethics, Fairness & Society
- Nisheeth K. Vishnoi
- Lingxiao Huang
- Eirini Ntoutsi
- Wenbin Zhang
- Dana Pessach
- Erez Shmueli
- L. Elisa Celis
- Catherine E. Tucker
Citations in the corpus
- Man is to Computer Programmer as Woman is to Homemaker? Debiasing Word Embeddings (2016, background): “A number of online systems have been shown to exhibit various biases, such as racial discrimination and gender bias in the ads presented to users [36, 6].”
- Examining Gender and Race Bias in Two Hundred Sentiment Analysis Systems (2018, background): “In general, any predictive model built on historical data may inadvertently inherit human biases based on gender, ethnicity, race, or religion (Sweeney, 2013; Datta et al., 2015).”
- Inherent Trade-Offs in the Fair Determination of Risk Scores (2016, background): “Second, in a very different domain, researchers have begun to analyze the ways in which different genders and racial groups experience advertising and commercial content on the Internet differently [9, 26].”
- How the machine ‘thinks’: Understanding opacity in machine learning algorithms (2016, background): “Some solutions perhaps wisely abandon answering the ‘why’ question and devise metrics that can, in other ways, evaluate discrimination (i.e. Datta et al., 2015).”
- Social Data: Biases, Methodological Pitfalls, and Ethical Boundaries (2019, background): “Others are designed by third party researchers to understand and audit system behavior (Datta et al., 2015).”
- Stable and Fair Classification (2019, background): “Fairness is often desired in recommendation systems, e.g., to check gender inequality in recommending high-paying jobs [24, 19, 68].”
- Racial categories in machine learning (2018, background): “Discriminating ads have been delivered based on racialized search results [51] and gendered user profiles [15].”
- Algorithmic Fairness (2020, background): “For instance, it was shown that Google’s ad-targeting algorithm had proposed higher-paying executive jobs more commonly for men than for women [41, 125].”
- Word Embeddings and Gender Stereotypes in Swedish and English (2019, background): “Online systems have been shown to exhibit a diverse range of biases and stereotypes, such as racism in ad delivery networks [41, 7] and exaggerated gender stereotypes in image search results [23].”
- LBS Research Online (2021, positive): “For example, Datta et al. (2015) states, ‘We cannot determine who caused these findings due to our limited visibility into the ad ecosystem, which includes Google, advertisers, websites, and users.”
- FAHT: An Adaptive Fairness-aware Decision Tree Classifier (2019, methodology): “…or even replacement of human-based decision making is widely adopted, growing concerns have been voiced from civil organizations, policy-makers and legal experts about the potential loss of fairness and accountability in the employed models [Datta et al. , 2015; Sweeney, 2013; Zliobaite, 2015].”
- A Qualitative Exploration of Perceptions of Algorithmic Fairness (2018, background): “To raise awareness and illustrate the potential for wide-ranging consequences, researchers and the press have pointed out a number of specific instances of algorithmic unfairness [19,58], for example, in predictive policing [19,43], the online housing marketplace [27,28], online ads [13,17,20,29,82]”
- The Ugly Truth About Ourselves and Our Robot Creations: The Problem of Bias and Social Inequity (2017, background): “For instance, a study by Datta et al. (2015) claims that Google displays far fewer ads for high-paying executive jobs to women than to men.”
- Detecting semantic anomalies (2019, background): “For a lot of existing datasets, it is quite possible to achieve good generalization performance without learning semantic distinctions, a possibility that spurs the search for removing algorithmic bias [35], and which is often exposed in embarrassing ways [36, 37].”
- Algorithmic Bias? An Empirical Study of Apparent Gender-Based Discrimination in the Display of STEM Career Ads (2019, background): “In settings where advertisements (ads) are allocated by algorithms, research has documented instances in which historically discriminated-against groups are more likely to be associated with undesirable ads (Sweeney 2013) and less likely to see desirable ads (Datta et al. 2015).”
- Fairness testing: testing software for discrimination (2017, background): “AdFisher computes a variant of group discrimination, but it could be integrated with Themis and its algorithms to measure causal discrimination.”
- Classification with Fairness Constraints: A Meta-Algorithm with Provable Guarantees (2018, background): “, against African-Americans while predicting future criminals [5, 7, 26], granting loans [17] or NYPD stopand-frisk [30], and against women while recommending jobs [16].”
- Constrained Learning With Non-Convex Losses (2021, neutral): “, adversarial examples) and unsafe behaviors [21]–[23].”