Direct Discrimination, Indirect Discrimination and Autonomy

Doyle, Oran

doi:10.1093/ojls/gqm008

Cited by 20 publications

(8 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It is evident within ECHR jurisprudence that both direct discrimination (read "disparate treatment" in North America labour law) and indirect discrimination (read "disparate impact" in North America labour law) are prohibited [83]. Although a heterogeneity of interpretations and applications of EU non-discrimination law exist [43,74], conceptually direct discrimination can be distilled as the use of an overtly prohibited ground of discrimination to unlawfully differentiate between persons or classes thereof in consequence in specified protected sectors [20,24,42,43,75]. Indirect discrimination entails the use of seemingly neutral criterion to do the same [20,24,43,75,83].…”

Section: Non-discrimination Law: Legal Definitionsmentioning

confidence: 99%

“…Although a heterogeneity of interpretations and applications of EU non-discrimination law exist [43,74], conceptually direct discrimination can be distilled as the use of an overtly prohibited ground of discrimination to unlawfully differentiate between persons or classes thereof in consequence in specified protected sectors [20,24,42,43,75]. Indirect discrimination entails the use of seemingly neutral criterion to do the same [20,24,43,75,83]. Intent is largely superfluous to these considerations [69].…”

Section: Non-discrimination Law: Legal Definitionsmentioning

confidence: 99%

See 1 more Smart Citation

Explainable artificial intelligence (XAI) post-hoc explainability methods: risks and limitations in non-discrimination law

2022

View full text Add to dashboard Cite

Organizations are increasingly employing complex black-box machine learning models in high-stakes decision-making. A popular approach to addressing the problem of opacity of black-box machine learning models is the use of post-hoc explainability methods. These methods approximate the logic of underlying machine learning models with the aim of explaining their internal workings, so that human examiners can understand them. In turn, it has been alluded that the insights from posthoc explainability methods can be used to help regulate black-box machine learning. This article examines the validity of these claims. By examining whether the insights derived from post-hoc explainability methods in post-model deployment can prima facie meet legal definitions in European (read European Union) non-discrimination law, we argue that machine learning post-hoc explanation methods cannot guarantee the insights they generate. Ultimately, we argue that the use of post-hoc explanatory methods is useful in many cases, but that these methods have limitations that prohibit reliance as the sole mechanism to guarantee fairness of model outcomes in high-stakes decision-making. By way of an ancillary function, the inadequacy of European Non-Discrimination Law for algorithmic decision-making is demonstrated too.

show abstract

Section: Non-discrimination Law: Legal Definitionsmentioning

confidence: 99%

Section: Non-discrimination Law: Legal Definitionsmentioning

confidence: 99%

Explainable artificial intelligence (XAI) post-hoc explainability methods: risks and limitations in non-discrimination law

2022

View full text Add to dashboard Cite

show abstract

“…Importantly, this definition captures and prohibits both direct and indirect discrimination. The difference between them is as follows (Doyle 2007):…”

Section: Other Articlesmentioning

confidence: 99%

Some Parents Are More Equal than Others: Discrimination against People with Disabilities under Adoption Law

Connell

2017

Laws

View full text Add to dashboard Cite

Article 23 of the Convention on the Rights of Persons with Disabilities (CRPD) explicitly includes 'the adoption of children' as a right to which people with disabilities are equally entitled. Despite the CRPD having been in force for over nine years, research is yet to consider whether CRPD signatory states have brought their respective adoption regimes in line with their obligations under art 23 of the CRPD. Using the laws of the Australian state of Victoria by way of case study, this article aims to shed light on the difficulties people with disabilities still face when attempting to adopt children. In terms of methodology, this article conducts an interpretive critique of Victoria's adoption law against art 23 of the CRPD, which it interprets mainly through the lens of the social model of disability. Ultimately, this article finds that Victoria's adoption framework closely resembles the adoption regimes of many other CRPD signatories, yet it clearly fails to uphold Australia's obligations under the CRPD. This is both as a result of the words of the legislation as well as their implementation in practice. This article proposes a suite of changes, both legislative and cultural, to bring Victoria's adoption framework in line with art 23, which it hopes will serve as a catalyst for change in other CRPD signatory states.

show abstract

“…Indirect discrimination is arguably the most likely type of discrimination to arise from automated decision making due to the reliance of these system on inference and proxies of target variables and protected attributes (Wachter, Mittelstadt, and Russell (2020)). Many scholars see value in judging discrimination with common sense (Doyle (2007)), however, this is often ineffective in cases of indirect discrimination, especially when the relation between the protected attribute and the neutral attribute is not straightforward (Wachter et al (2021)). Intuition might fail us because it cannot be assumed that automated systems will discriminate in ways similar to humans or follow their patterns of discrimination: new and counterintuitive proxies for traditionally protected attributes can emerge but will not necessarily be detected (Wachter et al (2021)).…”

Section: Introductionmentioning

confidence: 99%

PreCoF: Counterfactual Explanations for Fairness

Goethals

Martens

Calders

2022

Preprint

View full text Add to dashboard Cite

This paper studies how counterfactual explanations can be used to assess the fairness of a model. Using machine learning for high-stakes decisions is a threat to fairness as these models can amplify bias present in the dataset, and there is no consensus on a universal metric to detect this. The appropriate metric and method to tackle the bias in a dataset will be case-dependent, and it requires insight into the nature of the bias first. We aim to provide this insight by integrating explainable AI (XAI) research with the fairness domain. More specifically, apart from being able to use (Predictive) Counterfactual Explanations to detect explicit bias when the model is directly using the sensitive attribute, we show that it can also be used to detect implicit bias when the model does not use the sensitive attribute directly but does use other correlated attributes leading to a substantial disadvantage for a protected group. We call this metric $PreCoF$, or Predictive Counterfactual Fairness. Our experimental results show that our metric succeeds in detecting occurrences of implicit bias in the model by assessing which attributes are more present in the explanations of the protected group compared to the unprotected group. These results could help policymakers decide on whether this discrimination is justified or not.

show abstract

Direct Discrimination, Indirect Discrimination and Autonomy

Cited by 20 publications

References 0 publications

Explainable artificial intelligence (XAI) post-hoc explainability methods: risks and limitations in non-discrimination law

Explainable artificial intelligence (XAI) post-hoc explainability methods: risks and limitations in non-discrimination law

Some Parents Are More Equal than Others: Discrimination against People with Disabilities under Adoption Law

PreCoF: Counterfactual Explanations for Fairness

Contact Info

Product

Resources

About