Describe the ethical principles of trustworthy AI.

## Scenario (Applied) Your team is deploying a generative AI system to predict student performance and flag learners at risk of course failure. The model achieves 87% accuracy in internal tests. During a design review, a colleague proposes launching with predictions only, arguing that showing confidence scores or feature explanations will confuse instructors and slow adoption. Another engineer warns that the training data may reflect historical biases related to socioeconomic background, and that deploying without transparency mechanisms could lead to unfair interventions or privacy concerns [^5]. The **OECD AI Principles** emphasize that **trustworthy AI** must respect **human rights** and democratic values [^4], and research on similar educational AI systems highlights risks including unacceptable discrimination and breaches of privacy when vulnerable individuals' data is analyzed [^5]. You must decide whether to ship the prediction-only version now or delay the release to add explainability and bias-monitoring features. **What would you do, and why?**