Operationalizing Responsible AI
Abstract
Today we are going to look at ways to Operationalize Responsible AIso that we are building trustworthy systems that support mission needs. Imagine using an AI-enabled virtual reality headset for the first time at work. This headset is able to ask you questions and integrate your answers into the experience. You might expect to be asked basic questions about the purpose for your use, etc. Since this is a work device, consider what might feel too personal to share with the device. Next imagine the second time you put it on what would you expect that experience to be like? Would you expect the system to remember your preferences? To make a Responsible AI system, prior to designing those questions, someone needs to make decisions about what data are collected, how they are stored, and who has access to it. If someone accidentally provides access to their social security card or protected customer information, that information would need to be removed from the system. Additionally, the system would need to work for the different people in the organization they likely have different head sizes and heights, and would vary in their eye movements and eye colors. For example, someone with very light blue eyes is more sensitive to bright lights than someone with darker eyes.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 01, 2023
- Accession Number
- AD1214422
Entities
Organizations
- Carnegie Mellon University