电信系统与管理

Making AI Safer through identification and remediation of AI Risks Across Security, Ethics, Transparency, Compliance, and Other Pillars

Abstract

Murai Rao & Kashyap Murali

100 Billion Dollars. That’s how much AI spending is projected to rise to in a mere 3 years. With such rapid investment in a next-generational technology, its safety becomes of heightening concern to everyone that comes in touch with it from customers to vendors alike. Questions about the AI models safety across any potential security risks, particular bias towards a minority group, a lack of transparency in the decision-making process, or compliance violations from the AI environment are raised, and don’t have a proper answer. These questions don’t just apply to a specific vertical but it’s domain-agnostic and ensuring that a businesses AI is safe is what will set them apart from their customers. This talk will explore what AI safety is, why we need it, and how to ensure that one’s AI model is safe both technically and procedurally.