In AI-based systems, particularly those where theinternal structure has been generated by another software system, the complexity often makes it difficult for human testers to analyze the inner workings. As per the ISTQB Certified Tester AI Testing (CT-AI) Syllabus:
* Black-box testingis particularly useful when dealing with AI systems that have been generated by another system because:
* It allows testingwithout requiring knowledge of the internal logic.
* The AI model may be too complex for human testers to comprehend, making white-box testing ineffective.
* Black-box testing evaluates theinputs and outputs, ensuring functional correctnesswithout needing insight into how the system reaches a decision.
* Why other options are incorrect?
* A (Test automation and black-box testing): While automation is possible,black-box testing is not primarily about automationbut aboutabstracting the internal complexity.
* B (Understanding the logic of the software): This contradicts the premise of black-box testing, which is designed totest functionality without needing to understandthe inner workings.
* C (Checking transparency of the algorithm):Black-box testing does not check algorithm transparency-that would requirewhite-box testing or explainability techniques.
Thus, the best choice isOption D, as black-box testingremoves the need to analyze the internal structure of AI systems, making it the most appropriate testing method in this case.
Certified Tester AI Testing Study Guide References:
* ISTQB CT-AI Syllabus v1.0, Section 8.5 (Challenges Testing Complex AI-Based Systems)
* ISTQB CT-AI Syllabus v1.0, Section 8.6 (Testing the Transparency, Interpretability, and Explainability of AI-Based Systems)