Actionable Ethics for Fairness in AI
Abstract
Artificial intelligence (AI) is an increasingly pervasive tool with unintended consequences. AI technology and the tools that rely on it have been growing in use in all facets of society. From consumer uses such as whole-home automation systems to applications in manufacturing, space, and national security, AI is contributing to operational efficiencies. However, there is also growing awareness of AI solutions that have resulted in disparate impacts and harm. As AI becomes more ubiquitous, concerns about how and when it is used take on greater urgency and importance. These concerns should not sideline AI tools, which offer potential benefits. Rather, these concerns should be acknowledged and addressed by actively working to build fairness and ethical logic into AI solutions and by stress-testing these solutions for potential unintended consequences. This requires organizations to institutionalize "proactive ethics" into the AI lifecycle, which includes analyzing the ethical implications for stakeholders at all levels: individuals, groups, society, and the organization and its employees. Moving forward with ethical uncertainty poses harm to all stakeholders. Addressing the AI ethical challenge now is critical for establishing AI as a vehicle for creating a safer, more efficient, and more equitable world.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 2020
- Accession Number
- AD1137492
Entities
People
- Cathy Petrozzino
- Stuart Shapiro
Organizations
- MITRE Corporation