What is a recommended practice for monitoring AI systems?

• A. Set up effective logging, alerting, and monitoring for output quality
• B. Disable all logs to protect privacy
• C. Monitor only after issues occur
• D. Rely on random user feedback

Answer: (A)

Monitoring AI systems requires visibility into how the model behaves in production. A recommended practice is to set up effective logging, alerting, and ongoing monitoring focused on output quality. This provides continuous insight into when and why results may be off, helps detect data drift or model degradation early, and supports rapid remediation. It also creates an auditable record for governance and accountability, with alerts that prompt timely responses and, when appropriate, automated safeguards. Keep privacy and security in mind by logging only what’s necessary, redacting sensitive data, and enforcing strict access controls. Disabling logs undermines troubleshooting and accountability; waiting to monitor only after issues arise is reactive and risks missing root causes; relying on random user feedback is unreliable and biased.