Out of Context: Investigating the Bias and Fairness Concerns of "Artificial Intelligence as a Service"
arXiv:2302.01448 · doi:10.1145/3544548.3581463
Abstract
"AI as a Service" (AIaaS) is a rapidly growing market, offering various plug-and-play AI services and tools. AIaaS enables its customers (users) - who may lack the expertise, data, and/or resources to develop their own systems - to easily build and integrate AI capabilities into their applications. Yet, it is known that AI systems can encapsulate biases and inequalities that can have societal impact. This paper argues that the context-sensitive nature of fairness is often incompatible with AIaaS' 'one-size-fits-all' approach, leading to issues and tensions. Specifically, we review and systematise the AIaaS space by proposing a taxonomy of AI services based on the levels of autonomy afforded to the user. We then critically examine the different categories of AIaaS, outlining how these services can lead to biases or be otherwise harmful in the context of end-user applications. In doing so, we seek to draw research attention to the challenges of this emerging area.
arXiv commentAccepted to CHI '23: ACM Human Factors in Computing, 2023, Hamburg, Germany
References in corpus (32)
- Artificial Intelligence: the global landscape of ethics guidelines
- Model Cards for Model Reporting
- Equality of Opportunity in Supervised Learning
- Mitigating Bias in Algorithmic Hiring: Evaluating Claims and Practices
- Improving fairness in machine learning systems: What do industry practitioners need?
- 'It's Reducing a Human Being to a Percentage'; Perceptions of Justice in Algorithmic Decisions
- Fairness Beyond Disparate Treatment & Disparate Impact: Learning Classification without Disparate Mistreatment
- Expanding Explainability: Towards Social Transparency in AI systems
- Where Responsible AI meets Reality: Practitioner Perspectives on Enablers for shifting Organizational Practices
- Human-AI Collaboration in Data Science: Exploring Data Scientists' Perceptions of Automated AI
- Fairness in Learning: Classic and Contextual Bandits
- No Classification without Representation: Assessing Geodiversity Issues in Open Data Sets for the Developing World
- Closing the AI Accountability Gap: Defining an End-to-End Framework for Internal Algorithmic Auditing
- Everyday algorithm auditing: Understanding the power of everyday users in surfacing harmful algorithmic behaviors
- Decision Provenance: Harnessing data flow for accountable systems
- The De-democratization of AI: Deep Learning and the Compute Divide in Artificial Intelligence Research
- This Thing Called Fairness: Disciplinary Confusion Realizing a Value in Technology
- Does Object Recognition Work for Everyone?
- ATMSeer: Increasing Transparency and Controllability in Automated Machine Learning
- Exploring How Machine Learning Practitioners (Try To) Use Fairness Toolkits
- One button machine for automating feature engineering in relational databases
- Saving Face: Investigating the Ethical Concerns of Facial Recognition Auditing
- AutoAIViz: Opening the Blackbox of Automated Artificial Intelligence with Conditional Parallel Coordinates
- Minimax Pareto Fairness: A Multi Objective Perspective
- Calibrated Fairness in Bandits
- One Label, One Billion Faces: Usage and Consistency of Racial Categories in Computer Vision
- Image Counterfactual Sensitivity Analysis for Detecting Unintended Bias
- Promoting Fairness through Hyperparameter Optimization
- Reviewable Automated Decision-Making: A Framework for Accountable Algorithmic Systems
- Multi-Objective Automatic Machine Learning with AutoxgboostMC
- A Bandit-Based Algorithm for Fairness-Aware Hyperparameter Optimization
- FairAutoML: Embracing Unfairness Mitigation in AutoML