AI-900 Describe Artificial Intelligence workloads and considerations Practice Questions

Explanation: Fairness. This principle directly addresses AI systems that treat all people equitably and without bias. The translation system has significantly lower accuracy for Swahili and Navajo, which are underrepresented in the training data, creating an unfair bias against speakers of those languages. The goal of ensuring equitable service calls for applying the fairness principle to reduce this disparity.

Explanation: The scenario of avoiding discrimination based on age or disability is a fairness concern, which in Microsoft's responsible AI framework is addressed by the Transparency principle. Transparency ensures that AI systems are explainable and understandable, allowing stakeholders to audit decisions for bias and unfair outcomes. By making the prioritization process transparent, the hospital can identify and mitigate discriminatory patterns, promoting fairness indirectly through accountability and oversight.

Explanation: The scenario describes a model whose internal reasoning cannot be explained by the development team. This lack of explainability makes it impossible to verify whether the model is making unbiased decisions, directly violating the Microsoft responsible AI principle of Fairness. Fairness requires that AI systems treat all people fairly and avoid discrimination. Without understanding how the model reaches its conclusions, the organization cannot ensure that it is not unfairly denying loans based on protected attributes.

Explanation: Regression is the correct AI workload type because it is specifically designed to predict a continuous numeric value (e.g., price, temperature, sales amount) based on input features. Unlike classification, which predicts discrete categories, regression models output a real number by learning a mapping function from the input variables to a continuous target variable, often using algorithms like linear regression, decision trees, or neural networks.

Explanation: Sentiment analysis uses NLP to determine emotional tone and classify text as positive, negative, or neutral. In business, it is commonly used to monitor brand and product perception from customer feedback, making option B correct. Option D describes machine translation, not sentiment analysis.