Microsoft Azure AI Fundamentals (AI-900)Describe AI workloads and considerationsMedium
A company is using an AI model to predict customer churn. The model has achieved high accuracy in testing. However, stakeholders are concerned that they cannot understand *why* the model makes certain predictions, making it difficult to trust or explain decisions to customers. Which Responsible AI principle is most directly being violated or neglected?
- APrivacy and Security
- BTransparency
- CAccountability
- DFairness
Show answer & explanationAnswer & explanation
Correct answer: B. Transparency
Transparency (or explainability) is the principle that emphasizes the need to understand how an AI system arrives at its decisions. The inability to understand 'why' the model predicts churn directly indicates a lack of transparency.
Why the other options are wrong
- A. Privacy and Security concerns data protection, which is not the issue highlighted in the scenario.
- C. Accountability focuses on assigning responsibility for AI outcomes, not understanding the decision-making process itself.
- D. Fairness addresses biased outcomes, which is not the primary concern described here (the concern is *understanding* the decision).
Responsible AI: Transparency
The principle that AI systems should be understandable, allowing users and stakeholders to comprehend their decision-making processes.
- Aims for explainability and interpretability of AI models.
- Helps build trust and allows for debugging/improvement.
- Crucial for critical applications where 'why' matters.
Memory trick: AI should be TRANSPARENT, like clear glass, so you see through it.