Which of the following examples is a limited-risk AI application?

Limited-risk AI applications are systems that pose lower stakes for individuals and society, meaning their potential negative impact is manageable or less severe compared to other AI systems. Chatbots used for customer service fall into this category because they typically handle routine inquiries or provide assistance in a controlled environment. Their interactions do not usually involve high-stakes decision-making that could significantly affect a person's life, privacy, or rights.

In contrast, real-time biometric identification systems can result in privacy violations and discrimination if misapplied. A system assessing asylum eligibility has significant implications for individuals seeking refuge, affecting their safety and legal status, which makes it a high-risk application. AI systems in medical diagnostics also carry high stakes, as incorrect diagnoses can lead to severe consequences for individuals' health and well-being. Therefore, chatbots, with their relatively lower risk, exemplify a limited-risk AI application effectively.