Should We Audit AI Before We Adopt It?
Whenever we hear AI Audit, what comes to mind is a technology already in use. We examine its outputs for bias, test its accuracy, investigate failures, or study its effects on users. By then, however, an institution may already have purchased the technology, integrated it into existing systems, trained employees to use it, and introduced it to students or community members. What would change if some of this scrutiny occurred earlier? As AI-powered technologies become more common in schools, universities, libraries, and other information environments, there is a case for treating the period before adoption as an important point of inquiry. Since we increasingly audit what AI does, should we also audit what we know about an AI system before we decide to adopt it?
Read More
