The Fallacy of AI Functionality

Abstract

Deployed AI systems often do not work. They can be constructed haphazardly, deployed indiscriminately, and promoted deceptively. However, despite this reality, scholars, the press, and policymakers pay too little attention to functionality. This leads to technical and policy solutions focused on "ethical" or value-aligned deployments, often skipping over the prior question of whether a given system functions, or provides any benefits at all. To describe the harms of various types of functionality failures, we analyze a set of case studies to create a taxonomy of known AI functionality issues. We then point to policy and organizational responses that are often overlooked and become more readily available once functionality is drawn into focus. We argue that functionality is a meaningful AI policy challenge, operating as a necessary first step towards protecting affected communities from algorithmic harm.

References in corpus (23)

• Explaining and Harnessing Adversarial Examples
• Artificial Intelligence: the global landscape of ethics guidelines
• Common pitfalls and recommendations for using machine learning to detect and prognosticate for COVID-19 using chest radiographs and CT scans
• Mitigating Bias in Algorithmic Hiring: Evaluating Claims and Practices
• Demystifying the Draft EU Artificial Intelligence Act
• Prediction-Based Decisions and Fairness: A Catalogue of Choices, Assumptions, and Definitions
• The importance of transparency and reproducibility in artificial intelligence research
• The Malicious Use of Artificial Intelligence: Forecasting, Prevention, and Mitigation
• Release Strategies and the Social Impacts of Language Models
• Measurement and Fairness
• Problem Formulation and Fairness
• The Hidden Assumptions Behind Counterfactual Explanations and Principal Reasons
• Trust in Data Science: Collaboration, Translation, and Accountability in Corporate Data Science Projects
• Measuring Robustness to Natural Distribution Shifts in Image Classification
• Closing the AI Accountability Gap: Defining an End-to-End Framework for Internal Algorithmic Auditing
• Runaway Feedback Loops in Predictive Policing
• The De-democratization of AI: Deep Learning and the Compute Divide in Artificial Intelligence Research
• The effect of differential victim crime reporting on predictive policing systems
• ABOUT ML: Annotation and Benchmarking on Understanding and Transparency of Machine Learning Lifecycles
• MLSys: The New Frontier of Machine Learning Systems
• Beyond Near- and Long-Term: Towards a Clearer Account of Research Priorities in AI Ethics and Society
• Measurement as governance in and for responsible AI
• Engagement Outweighs Exposure to Partisan and Unreliable News within Google Search

Cited by in corpus (33)

• Auditing large language models: a three-layered approach
• Auditing of AI: Legal, Ethical and Technical Approaches
• Harms from Increasingly Agentic Algorithmic Systems
• ELEPHANT: Measuring and understanding social sycophancy in LLMs
• The Dark Side of AI Companionship: A Taxonomy of Harmful Algorithmic Behaviors in Human-AI Relationships
• The Situate AI Guidebook: Co-Designing a Toolkit to Support Multi-Stakeholder Early-stage Deliberations Around Public Sector AI Proposals
• Not My Voice! A Taxonomy of Ethical and Safety Harms of Speech Generators
• Null Compliance: NYC Local Law 144 and the Challenges of Algorithm Accountability
• Three lines of defense against risks from AI
• Towards AI Accountability Infrastructure: Gaps and Opportunities in AI Audit Tooling
• Sketching AI Concepts with Capabilities and Examples: AI Innovation in the Intensive Care Unit
• Are We Asking the Right Questions?: Designing for Community Stakeholders' Interactions with AI in Policing
• The Dark Side of Dataset Scaling: Evaluating Racial Classification in Multimodal Models
• Toward Operationalizing Pipeline-aware ML Fairness: A Research Agenda for Developing Practical Guidelines and Tools
• The Fall of an Algorithm: Characterizing the Dynamics Toward Abandonment
• Mapping the individual, social, and biospheric impacts of Foundation Models
• ARTiST: Automated Text Simplification for Task Guidance in Augmented Reality
• Epistemic Power in AI Ethics Labor: Legitimizing Located Complaints
• LLM-Driven Robots Risk Enacting Discrimination, Violence, and Unlawful Actions
• The "Who", "What", and "How" of Responsible AI Governance: A Systematic Review and Meta-Analysis of (Actor, Stage)-Specific Tools
• Counterfactual Prediction Under Outcome Measurement Error
• A Human-Centered Review of Algorithms in Homelessness Research
• Legacy Procurement Practices Shape How U.S. Cities Govern AI: Understanding Government Employees' Practices, Challenges, and Needs
• Making the Right Thing: Bridging HCI and Responsible AI in Early-Stage AI Concept Selection
• Who is Responsible When AI Fails? Mapping Causes, Entities, and Consequences of AI Privacy and Ethical Incidents
• The Datafication of Care in Public Homelessness Services
• One Bad NOFO? AI Governance in Federal Grantmaking
• From Model Performance to Claim: How a Change of Focus in Machine Learning Replicability Can Help Bridge the Responsibility Gap
• Investigating Youth's Technical and Ethical Understanding of Generative Language Models When Engaging in Construction and Deconstruction Activities
• Better Together? The Role of Explanations in Supporting Novices in Individual and Collective Deliberations about AI
• Are machine learning technologies ready to be used for humanitarian work and development?
• Toward Substantive Intersectional Algorithmic Fairness: Desiderata for a Feminist Approach
• Implicit Humanization in Everyday LLM Moral Judgments