Identify practical use cases for AI

Computer Vision (Option B) is required to process the live camera feed and visually identify defects or damage on physical packages. Time-series forecasting (Option C) is required to analyze historical shipping volumes over time to predict future numerical demand, informing staffing decisions. NLP (Option A) and Sentiment Analysis (Option D) process human language and emotions, which are irrelevant to package damage and shipping volume. Generative AI (Option E) synthesizes new content rather than forecasting operational metrics or classifying physical defects. Reference: https://docs.aws.amazon.com/rekognition/latest/dg/what-is.html

Reinforcement learning (Option C) adds massive value to logistics optimization because it trains an autonomous agent to achieve a goal (minimizing fuel) through trial and error in an environment, driven by explicit rewards and penalties. Sentiment analysis (Option A) evaluates emotional tone in text. Generative AI (Option B) creates novel content. Unsupervised clustering (Option D) groups unlabeled data but does not autonomously adapt routing actions based on environmental feedback. Reference: https://docs.aws.amazon.com/sagemaker/latest/dg/reinforcement-learning.html

Time-Series Forecasting (Option A) is required to predict future continuous values based strictly on chronological sequencing (e.g., daily sales trends). Clustering (Option C) is required to segment an unlabeled customer base into cohorts using unsupervised learning. Regression (Option E) is required to estimate the exact continuous numerical dollar amount for the customer's LTV. Binary Classification (Option B) is incorrect because predicting LTV requires forecasting a continuous dollar amount, not a discrete 'Yes/No' outcome. Reinforcement Learning (Option D) is for autonomous decision-making agents, not forecasting or clustering. Reference: https://docs.aws.amazon.com/forecast/latest/dg/what-is-forecast.html

Amazon Translate (Option B) is required to convert the foreign language hotel reviews into English in real-time using neural machine translation. Amazon Comprehend (Option E) is required to then scan the English text using natural language processing to extract entities and key phrases (like 'clean room'). Amazon Lex (Option A) is used to build chatbots, not process static review text. Amazon Transcribe (Option C) converts speech to text, but reviews are already written. Amazon Polly (Option D) converts text into spoken audio, which is not required. Reference: https://docs.aws.amazon.com/translate/latest/dg/what-is.html

Amazon Comprehend (Option D) is an NLP service that uses machine learning to uncover insights and relationships in text, natively providing both sentiment analysis (positive/negative/neutral) and entity extraction. Amazon Translate (Option A) converts text from one language to another but does not extract sentiment. Amazon SageMaker (Option B) is a platform to build custom models, which is overkill since Comprehend provides this out of the box. Amazon Textract (Option C) extracts text and data from scanned physical documents (like PDFs or images), rather than analyzing the sentiment of digital emails. Reference: https://docs.aws.amazon.com/comprehend/latest/dg/how-it-works.html

The task requires deterministic execution (Option A). A mathematical formula guarantees a 100% accurate, predictable outcome every time using standard rule-based programming (IF/THEN logic). Machine learning models generate probabilistic predictions, which are unacceptable when exact legal calculations are required. AI models process numerical data exceptionally well, so Option B is false. Supervised learning does not strictly require real-time streams (Option C is false). Generative AI is for creating novel content, not tabular arithmetic (Option D is false). Reference: https://aws.amazon.com/machine-learning/what-is-machine-learning/

Clustering (Option C) is the correct technique. Topic modeling is a form of unsupervised clustering applied to text, allowing algorithms (like LDA or Neural Topic Models in SageMaker) to group documents with similar semantic themes without predefined labels. Linear Regression (Option A) predicts continuous numerical values. Logistic Regression (Option B) is a classification algorithm that requires labeled data to assign categories. Reinforcement Learning (Option D) trains agents via reward loops in dynamic environments, which is entirely unrelated to grouping text documents. Reference: https://docs.aws.amazon.com/sagemaker/latest/dg/ntm.html

A traditional ML model (Option B) is required. Algorithms like logistic regression or decision trees offer high explainability (often referred to as 'white box' models), allowing engineers to explicitly demonstrate how individual input features (income, debt) mathematically impact the final decision. Foundation Models (Option A and C) are massively complex neural networks with billions of parameters that function as 'black boxes', making it nearly impossible to extract the exact weight of a single factor to satisfy strict legal audits. Both models can be run securely in isolated cloud environments (Option D is false). Reference: https://docs.aws.amazon.com/machine-learning/latest/dg/machine-learning.html