Work

Advancing AI Audits for Enhanced AI Governance

Year: 2023
Type: report
Abstract: As artificial intelligence (AI) is integrated into various services and systems in society, many companies and organizations have proposed AI principles, policies, and made the related commitments.Conversely, some have proposed the need for independent audits, arguing that the voluntary principles adopted by the developers and providers of AI services and systems insufficiently address risk.This policy recommendation summarizes the issues related to the auditing of AI services and systems and presents a vision for AI auditing that contributes to sound AI governance.The issues surrounding AI auditing are diverse, and even if the same word is used, different assumptions are made based on different positions and preconditions, making it easy for discussions to be at odds.Therefore, it's important to have common understanding when the parties discuss issues such as audit scope and the timing for putting AI audits into practice.This policy recommendation addresses the following six issues that are assumed when discussing AI auditing and these are presented in Chapter 2: (1) The need for AI audits, (2) proof proposition, (3) AI audit scope, (4) timing of AI audits, (5) AI audit practitioner requirements, and (6) AI auditing parties and organizations involved.Chapter 3 examines why AI audits are difficult to conduct from variety of perspective like technical, institutional, and social perspective.Chapter 4 presents the case of recruitment AI to provide specific examples of issues related to AI audits and the reasons why AI audits are difficult to conduct.In Chapter 5, three recommendations for promoting AI auditing are explained. Recommendation 1: Development of institutional design for AI audits Recommendation 2: Training human resources for AI audits Recommendation 3: Updating AI audits in accordance with technological progressIn this policy recommendation, AI is assumed to be that which recognizes and predicts data with the last chapter outlining how generative AI should be audited.
Language: English
Open Access status: green
OpenAlex ID: W7146651299
Cites: 7
Cited by:
Related to: 10