Testing Autonomous AI-Based Systems CT-AI Practice Questions (Page 2)

Questions 6–10

1. 6application · medium

   An autonomous shuttle service is being deployed in a European city. Before full public operation, the company must demonstrate safe operation in real traffic. Regulatory approval requires evidence of safe behavior in mixed traffic with pedestrians and cyclists. The company has already completed extensive simulation testing. What is the most appropriate next step to gather real-world evidence while managing risk?

   • APerform additional simulation testing with higher-fidelity pedestrian models to eliminate the need for real-world testing.
   • BRun a phased field trial on designated public roads with a safety driver onboard, gradually expanding the operational design domain based on collected data.
   • CConduct a supervised field trial on a closed test track that replicates city traffic with professional actors playing pedestrians and cyclists.
   • DLaunch the full public service immediately and rely on operational monitoring to detect and correct any safety issues.
   Reveal Solution Discussion Select an answer first
2. 7foundation · easy

   What is the primary goal of risk assessment in the context of autonomous AI systems?

   • ATo ensure the system is profitable
   • BTo test the system's response to random inputs
   • CTo guarantee that the system will never cause harm
   • DTo determine the likelihood and severity of potential harms
   Reveal Solution Discussion Select an answer first
4. 8expert · hard

   An autonomous delivery robot operates on public sidewalks. During simulation testing, the team discovers that the robot occasionally misclassifies a child's toy as a small animal and performs a sudden stop. The sudden stop has a 1% chance of causing a pedestrian behind the robot to trip. The team must decide whether to deploy the robot. The robot's task is to deliver packages, and the misclassification occurs in 0.5% of encounters with toys. Which risk assessment approach is most appropriate for this decision?

   • ADeploy the robot because the probability of a trip is only 0.005% (0.5% × 1%), which is below the company's acceptable risk threshold of 0.01%.
   • BDeploy the robot but add a warning sound before sudden stops to reduce the risk of tripping, without further analysis.
   • CDo not deploy the robot because any risk of injury to a pedestrian is unacceptable, regardless of probability.
   • DConduct a formal risk assessment that combines the probability of the trip with the severity of the resulting injury, and compare the resulting risk level against the company's risk acceptance criteria.
   Reveal Solution Discussion Select an answer first
5. 9foundation · easy

   How do simulation environments enable testing of autonomous systems 'at scale'?

   • ABy using larger physical test tracks
   • BBy testing only the most common scenarios
   • CBy requiring more test drivers
   • DBy allowing many test scenarios to be run quickly and in parallel
   Reveal Solution Discussion Select an answer first
6. 10foundation · easy

   What is the role of operational monitoring in the testing of autonomous systems?

   • ATo monitor the system's internal code for syntax errors
   • BTo continuously assess system behavior during real-world operation to detect anomalies
   • CTo replace the need for pre-deployment testing
   • DTo ensure the system is used only by authorized personnel
   Reveal Solution Discussion Select an answer first