Explaining models an empirical study of how explanations impact fairness judgment

By D. Jonathan, Liao Q. Vera, Yunfeng Zhang, K. Rachel, D. Casey. 2019

346 citations recorded by the source corpus.

How researchers see this paper

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

1. Vivian Lai
2. Q. Liao
3. Chenhao Tan
4. Xinru Wang
5. Bo Li
6. Chacha Chen
7. Longo Luca
8. Jennifer Wortman Vaughan

Citations in the corpus

• In AI We Trust? Factors That Influence Trustworthiness of AI-infused Decision-Making Processes (2019, methodology): “Similar scenario-based designs have been used to evaluate peoples’ perceptions of AI (e.g., [9]) and IoT (e.g., [39]) systems.”
• Manipulating and Measuring Model Interpretability (2018, background): “[29] studied the effects of different types of explanations on people’s perceptions of a model’s fairness; and Kaur et al.”
• Effect of confidence and explanation on accuracy and trust calibration in AI-assisted decision making (2020, methodology): “calibration has been studied for automation [13, 18, 20, 26], as to prevent both automation aversion and automation bias, it is not well understood in the context of AI systems. In one relevant study [8], Dodge et al. compared the effect of different explanation methods for calibrating perceived fai”
• Human-Centered Explainable AI (XAI): From Algorithms to User Experiences (2021, methodology): “When it comes to auditing for model fairness, our work with collaborators compared the effectiveness of four types of explanation (shown in Figure 2) and found that contrastive explanations can effectively help people identify concerns of individual fairness, where similar individuals from different”
• Towards a Science of Human-AI Decision Making: A Survey of Empirical Studies (2021, methodology): “Model uncertainty Classification confidence (or probability) [10, 13, 21, 22, 43, 60, 68, 73, 86, 87, 144, 155], flag low confidence [90], uncertainty distribution (regression) [101] Local feature importance Coefficients [16, 28, 39, 43, 49, 55, 80, 81, 86, 87, 94, 106, 111, 125, 128, 143], attentio”
• Algorithmic bias: review, synthesis, and future research directions (2021, background): “Some of those studies, however, have reported contradictory results regarding effective explanation styles (Binns et al., 2018; Dodge et al., 2019) and impacts of machine advice on decisions (Grgić-Hlača et al., 2019; Rhue, 2019), highlighting the need for additional research to better understand…”
• Counterfactual Explanations and Algorithmic Recourses for Machine Learning: A Review (2020, unknown): “[81] performed user studies that showed that users prefer CFEs over case-based”
• Expanding Explainability: Towards Social Transparency in AI systems (2021, background): “Only until recently have researchers began to conduct controlled lab studies to rigorously evaluate popular XAI techniques [12, 13, 15, 22, 27, 53, 80], as well as studies to understand real-world user needs for AI explainability [43, 50, 58].”
• Trustworthy AI: From Principles to Practices (2021, background): “Furthermore, studies on fairness and explainability have shown several approaches to explanation to be unfair [98].”
• Are Explanations Helpful? A Comparative Study of the Effects of Explanations in AI-Assisted Decision-Making (2021, methodology): “This dataset was widely used by researchers to understand how people interact with machine assistance in their decision making [23, 31].”
• Counterfactual Explanations for Machine Learning: A Review (2020, background): “Binns et al. [18] and Dodge et al. [32] performed user-studies which showed that users prefer counterfactual explanations over case-based reasoning, which is another example-based approach.”
• Explainable Artificial Intelligence (XAI) 2.0: A Manifesto of Open Challenges and Interdisciplinary Research Directions (2023, background): “While some studies evaluated the impact of synthesized explanations of AI systems on humans when compared to the scenario where no explanations were provided [207–209], there is a need for more (and more systematic) work on the topic.”
• WeBuildAI (2019, background): “Other scholars systematically investigated perceived fairness of algorithmic decisions in hiring [47], recidivism [31], child welfare services [18], and resource allocation such as task assignment [48] and goods division [49].”