Unit 7: Generative AI

Correct answer: A. Dividing text or a sentence into smaller units such as words, subwords, or punctuation marks

Tokenization divides text into smaller units called tokens, such as words, subwords, or punctuation marks. This helps an NLP system process the text, so option A is correct. Translation, sentiment analysis, and dependency analysis are different NLP tasks. Tokenization is commonly used near the beginning of text preprocessing.

Correct answer: A. Clarifying the problem and its objective

Problem scoping defines the problem, its objective, boundaries, stakeholders, and expected outcome. This planning gives direction to later data collection, analysis, and model development. Deleting relevant data, hiding results, or using resources without a plan does not define a project and can reduce its reliability and accountability.

Correct answer: A. To present the main ideas of a long text concisely

Text summarization presents the essential information and main ideas of a longer text in fewer words while preserving its central meaning. A good summary may retain important numerical facts when they are relevant. Reversing words, converting images to sound, or removing all numbers is not summarization.

Correct answer: A. It helps users understand an AI system’s capabilities, limitations, and how it works.

Transparency means clearly communicating an AI system’s purpose, operation, capabilities, limitations, and possible risks. This information helps users judge when the system is suitable and interpret its outputs carefully. The other options describe the opposite of transparency.

Correct answer: A. To reduce risks and harmful outcomes

Safety testing helps identify errors, potential risks, and unintended harmful outcomes in an AI system before or during its use. This allows corrective measures to be taken and supports safer, more responsible deployment. Option B is incorrect because testing aims to reduce incorrect results, not increase them. Exam tip: In AI ethics, associate safety testing with risk control and harm prevention.

Correct answer: A. Answering questions and providing assistance

An AI chat assistant uses language-processing and, in generative systems, language-generation capabilities to interpret a user’s request and produce an informative response or practical guidance. Therefore, option A is the valid use. Deleting every question, intentionally confusing users, or printing without understanding language is not the normal purpose of a conversational assistant.