AWS AI Practitioner practice test

In prompt engineering, providing a few labeled examples within the prompt to guide a foundation model's response is known as:

A bank trains a model to predict whether a loan applicant will default, using historical data that includes the known outcome for each applicant. This is an example of which type of machine learning?

Which AWS capability helps detect bias in machine learning datasets and models and provides explainability reports for model predictions?

An organization wants its generative AI chatbot to answer questions using the company's internal documents while reducing hallucinations. The most appropriate technique is to:

When invoking a foundation model, lowering the temperature parameter toward zero will generally cause the model to produce output that is:

A company needs to automatically extract sentiment, key phrases, and named entities from large volumes of customer feedback text. Which AWS service is purpose built for this?

A company has a small labeled dataset specific to its industry and wants a foundation model to better follow its domain terminology. Compared with pre-training a model from scratch, fine-tuning is preferable because it:

A retailer wants to detect objects, scenes, and inappropriate content in user-uploaded images without building a custom model. Which AWS service should be used?

An AI system used for hiring is found to favor one demographic group over another due to patterns in its training data. This problem is best described as:

Which AWS service provides a fully managed environment to build, train, and deploy machine learning models at scale across the entire ML lifecycle?

When evaluating a classification model on an imbalanced dataset where only 2 percent of cases are positive, accuracy can be misleading because:

Which of the following is a typical use case well suited to a generative AI foundation model?

Which AWS service should be used to centrally manage fine-grained permissions that control who can invoke a foundation model in Amazon Bedrock?

A machine learning model performs extremely well on its training data but poorly on new, unseen data. This condition is known as:

Amazon Q Business is best described as a service that:

An attacker crafts input designed to make a foundation model ignore its original instructions and reveal restricted information. This security threat is known as:

Guardrails for Amazon Bedrock are primarily used to:

Vector embeddings are commonly stored in a vector database so that a generative AI application can:

When choosing a foundation model for a production application, which factor is most important to evaluate alongside accuracy?