Microsoft Azure AI Fundamentals (AI-900)Describe AI workloads and considerationsHard
A pharmaceutical company is using an AI model to predict the efficacy of new drug compounds. The model's predictions are highly complex, and it's difficult for human scientists to understand exactly why a particular compound is predicted to be effective or ineffective. The company wants to ensure that internal researchers can scrutinize the model's reasoning to validate its scientific basis. Which responsible AI principle is being emphasized here?
- APrivacy and Security
- BTransparency
- CInclusiveness
- DFairness
Show answer & explanationAnswer & explanation
Correct answer: B. Transparency
The challenge of understanding 'exactly why a particular compound is predicted to be effective or ineffective' and the desire to 'scutinize the model's reasoning' directly point to the principle of Transparency. This principle, often supported by Explainable AI (XAI) techniques, ensures that AI decisions are understandable and explainable to humans.
Why the other options are wrong
- A. Privacy and Security deal with data protection, not the explainability of the model's scientific predictions.
- C. Inclusiveness is about broad accessibility for users, not the interpretability of a complex model's scientific reasoning.
- D. Fairness focuses on preventing bias against groups, not on understanding the model's internal logic for drug efficacy.
Responsible AI: Transparency
Ensuring AI systems are understandable, and their decision-making processes and rationale are explainable.
- Crucial for trust, auditing, and debugging complex AI models.
- Involves techniques like Explainable AI (XAI).
- Allows human experts to validate AI's reasoning.
Memory trick: AI's brain needs to be TRANSPARENT, like a clear window.