Jury Learning: Integrating Dissenting Voices into Machine Learning Models

Abstract

Whose labels should a machine learning (ML) algorithm learn to emulate? For ML tasks ranging from online comment toxicity to misinformation detection to medical diagnosis, different groups in society may have irreconcilable disagreements about ground truth labels. Supervised ML today resolves these label disagreements implicitly using majority vote, which overrides minority groups' labels. We introduce jury learning, a supervised ML approach that resolves these disagreements explicitly through the metaphor of a jury: defining which people or groups, in what proportion, determine the classifier's prediction. For example, a jury learning model for online toxicity might centrally feature women and Black jurors, who are commonly targets of online harassment. To enable jury learning, we contribute a deep learning architecture that models every annotator in a dataset, samples from annotators' models to populate the jury, then runs inference to classify. Our architecture enables juries that dynamically adapt their composition, explore counterfactuals, and visualize dissent.

arXiv commentTo appear at CHI 2022

References in corpus (8)

• Deep Learning with Differential Privacy
• Showing Academic Performance Predictions during Term Planning: Effects on Students' Decisions, Behaviors, and Preferences
• DCN V2: Improved Deep & Cross Network and Practical Lessons for Web-scale Learning to Rank Systems
• Society-in-the-Loop: Programming the Algorithmic Social Contract
• Roles for Computing in Social Change
• Do Datasets Have Politics? Disciplinary Values in Computer Vision Dataset Development
• Keeping Community in the Loop: Understanding Wikipedia Stakeholder Values for Machine Learning-Based Systems
• PolicyKit: Building Governance in Online Communities

Cited by in corpus (27)

• Collective Constitutional AI: Aligning a Language Model with Public Input
• Understanding Practices, Challenges, and Opportunities for User-Engaged Algorithm Auditing in Industry Practice
• Sensible AI: Re-imagining Interpretability and Explainability using Sensemaking Theory
• Toward a Perspectivist Turn in Ground Truthing for Predictive Computing
• Farsight: Fostering Responsible AI Awareness During AI Application Prototyping
• Value Kaleidoscope: Engaging AI with Pluralistic Human Values, Rights, and Duties
• Beyond Preferences in AI Alignment
• Model Sketching: Centering Concepts in Early-Stage Machine Learning Model Design
• Remove, Reduce, Inform: What Actions do People Want Social Media Platforms to Take on Potentially Misleading Content?
• Wikibench: Community-Driven Data Curation for AI Evaluation on Wikipedia
• Searching For or Reviewing Evidence Improves Crowdworkers' Misinformation Judgments and Reduces Partisan Bias
• DISCERN: Designing Decision Support Interfaces to Investigate the Complexities of Workplace Social Decision-Making With Line Managers
• Cura: Curation at Social Media Scale
• Plurals: A System for Guiding LLMs Via Simulated Social Ensembles
• Ontologies in Design: How Imagining a Tree Reveals Possibilites and Assumptions in Large Language Models
• The Ecological Fallacy in Annotation: Modelling Human Label Variation goes beyond Sociodemographics
• Auditing Cross-Cultural Consistency of Human-Annotated Labels for Recommendation Systems
• A Matter of Perspective(s): Contrasting Human and LLM Argumentation in Subjective Decision-Making on Subtle Sexism
• A General Model for Aggregating Annotations Across Simple, Complex, and Multi-Object Annotation Tasks
• Inclusive Portraits: Race-Aware Human-in-the-Loop Technology
• Wisdom of the Crowd, Without the Crowd: A Socratic LLM for Asynchronous Deliberation on Perspectivist Data
• Evaluating AI systems under uncertain ground truth: a case study in dermatology
• Architectural Sweet Spots for Modeling Human Label Variation by the Example of Argument Quality: It's Best to Relate Perspectives!
• Mitigating Voter Attribute Bias for Fair Opinion Aggregation
• The Chilling: Identifying Strategic Antisocial Behavior Online and Examining the Impact on Journalists
• Critically Engaged Pragmatism: Scientific Norm and Social, Pragmatist Epistemology for AI Science Evaluation Tools
• Perils of Label Indeterminacy: A Case Study on Prediction of Neurological Recovery After Cardiac Arrest