Leading experts at the VB AI Impact Tour in New York City, presented by UiPath, discussed the critical issue of auditing AI models for bias, performance, and ethical standards. Michael Raj of Verizon Communications, Rebecca Qian of Patronus AI, and Matt Turck of FirstMark shared their perspectives on methodologies and best practices. Justin Greenberger of UiPath emphasized the need for more frequent risk evaluations, pointing out the importance of understanding and mitigating risks on a nearly monthly basis. He also highlighted the growing importance of frameworks like those from the Institute of Internal Auditors (IIA) and the foundational role of regulations such as GDPR in data security.
As organizations pilot AI projects, they face challenges such as finding subject matter experts and educating employees about the scope and limitations of new technologies like generative AI. The process of integrating generative AI into existing workflows rather than overhauling entire systems presents additional audit challenges, particularly in monitoring the use of private data.
The role of humans in AI-driven processes is evolving. While human decision-making is still crucial, Greenberger predicts that as audit controls improve, human roles may shift towards more creative and emotional responsibilities, with decision-making increasingly handled by AI. This transition underscores the need for continuous adaptation and education for managers and executives.
Leading experts discuss the methodologies and challenges of auditing AI for bias, performance, and ethical standards.
