Adversarial Attacks CT-AI Practice Questions (Page 2)

Questions 6–10

1. 6foundation · easy

   What is the purpose of testing an AI system for adversarial robustness?

   • ATo identify vulnerabilities by subjecting the system to adversarial examples and measuring its performance.
   • BTo ensure the system meets legal and regulatory requirements for AI safety.
   • CTo compare the system's performance against other AI systems on the same dataset.
   • DTo determine how well the system performs on clean, unperturbed data.
   Reveal Solution Discussion Select an answer first
2. 7foundation · easy

   Why are adversarial attacks considered a significant concern for AI-based systems?

   • AThey are only a concern for AI systems that use deep learning, not for other types of machine learning models.
   • BThey can cause the AI system to make incorrect predictions, potentially leading to security breaches or safety hazards in critical applications.
   • CThey always result in the complete shutdown of the AI system, making it unavailable for legitimate users.
   • DThey are primarily a theoretical concern with no practical impact on deployed AI systems.
   Reveal Solution Discussion Select an answer first
4. 8foundation · easy

   In a security-sensitive application like facial recognition for access control, what is a serious security impact of an adversarial attack?

   • AThe system will become more accurate because it learns from the adversarial examples.
   • BThe system will require more computational resources to process the adversarial input.
   • CAn attacker could craft a pattern that causes the system to recognize them as an authorized user, bypassing security.
   • DThe system will automatically shut down to prevent further attacks.
   Reveal Solution Discussion Select an answer first
5. 9application · medium

   A company is deploying an AI-based system for medical diagnosis. They are concerned about adversarial attacks that could cause misdiagnosis. What is the primary reason to test for adversarial robustness?

   • ATo improve the model's accuracy on normal cases
   • BTo ensure patient safety by preventing misdiagnosis due to malicious inputs
   • CTo reduce the model's computational requirements
   • DTo make the model easier to interpret
   Reveal Solution Discussion Select an answer first
6. 10application · medium

   A tech company has an AI model that provides recommendations. A competitor wants to replicate the model's functionality by querying it with many inputs and observing the outputs. Which type of attack is this?

   • AExtraction attack
   • BPoisoning attack
   • CEvasion attack
   • DInference attack
   Reveal Solution Discussion Select an answer first