Watch this video on YouTube
Scenario: You are tasked with assessing the vulnerability of an AI system to adversarial attacks. What approach should you use to determine if small, carefully crafted changes to inputs can cause the system to fail?