Reacting to unforeseen events outside of the training environment describes which challenge?

• A. Unpredictability in AI
• B. Data governance
• C. ISACA COBIT
• D. Lagging indicators

Answer: (A)

Reacting to unforeseen events outside of the training environment describes unpredictability in AI. When a model encounters inputs or scenarios it didn’t see during training, its behavior can become unexpected or unsafe because it wasn’t taught how to handle those cases. This challenge comes from distribution shifts and out-of-distribution data, where real-world situations differ from the data the model was built on. To cope, systems need awareness of uncertainty, mechanisms to detect anomalies, and safeguards like fallback paths or human review, so they don’t rely on incorrect assumptions about what the model will do in new contexts. Data governance, COBIT, and lagging indicators address different concerns—data management rules, IT governance frameworks, and retrospective performance measurement, respectively—and don’t directly describe the issue of handling novel, unseen situations.