IBM A1000-078 - Assessment: Foundations of AI Practice Exam 2025: Latest Questions

Which statement best describes the difference between Artificial Intelligence (AI) and Machine Learning (ML)?

A data scientist wants to predict whether a customer will churn (yes/no) based on historical customer behavior. Which type of machine learning problem is this?

A team wants to quickly build a chatbot for answering common HR questions using a managed IBM Watson service with conversational features. Which service is most appropriate?

Which approach is an example of AI augmentation (assistive AI) rather than AI automation?

A model performs well on training data but significantly worse on new, unseen data. Which issue is most likely occurring?

A team is deploying an AI system for hiring recommendations. Which action best supports responsible AI and helps reduce unintended bias?

You have thousands of internal policy documents (PDFs, Word files) and need employees to ask natural-language questions and get answers grounded in those documents. Which IBM Watson service is best suited to index and retrieve insights from this content?

A retailer wants to group customers into segments based on purchasing behavior, but they do not have predefined segment labels. Which technique is most appropriate?

A company is building a support assistant that must (1) chat with users, and (2) answer detailed questions using information from a large set of product manuals. Which architecture best fits this requirement using IBM Watson services?

A fraud detection model is trained on last year's transaction patterns. After deployment, business processes and customer behavior change, and the model's fraud detection rate drops over time. What is the most likely cause and best next step?

A retail team wants a simple way to explain what “AI” means to non-technical stakeholders. Which statement best describes Artificial Intelligence at a foundational level?

A model predicts whether an email is “spam” or “not spam” based on labeled examples. Which type of machine learning is being used?

A customer support chatbot needs to identify a user’s intent (e.g., “reset password”, “track order”) and extract entities (e.g., order number). Which IBM Watson service is the best fit?

A team finds that their model performs extremely well on training data but significantly worse on new, unseen data. What is the most likely issue?

An organization wants to detect whether a deployed model’s input data characteristics have changed compared to training data, potentially reducing accuracy. Which concept best describes this problem?

A bank must provide governance for its AI models: monitor quality, detect bias, and explain outcomes. Which IBM capability is primarily used for these tasks?

A data scientist notices the training dataset has 98% “non-fraud” and 2% “fraud.” The model predicts almost everything as non-fraud and shows high accuracy but misses most fraud cases. Which metric is typically more appropriate to focus on for the fraud class?

A healthcare organization wants to search across thousands of clinical PDFs and extract relevant passages to answer clinician questions, while returning supporting evidence from the documents. Which Watson service is most suitable?

A city plans to deploy an AI solution that recommends police patrol locations. They are concerned about historical bias in crime data leading to unfair targeting of neighborhoods. What is the most appropriate first step before model training?

A team trains a model using customer data. During deployment, they cannot reproduce training results, and the model behavior differs between environments. Which practice most directly improves reproducibility and auditability?

A helpdesk team wants to quickly categorize incoming support tickets into predefined categories (e.g., Billing, Technical Issue, Account Access) using historical labeled tickets. Which machine learning approach best fits this problem?

A team is building an AI model to recommend which job applicants to interview. They notice the training data historically contains fewer successful applicants from a particular group due to past hiring practices. What is the most appropriate first step to reduce the risk of biased outcomes?

A company wants to build a chatbot for internal HR policies. The bot should provide grounded answers based on approved HR documents and must reduce the chance of hallucinated responses. Which architecture pattern is most appropriate?

A developer uses IBM Watson Natural Language Understanding to extract entities from customer emails. Results are inconsistent across emails with different formatting and lots of boilerplate signatures. What is the best troubleshooting step to improve extraction quality?

A data scientist trained a model with 95% accuracy. After deployment, performance drops sharply because real-world customer behavior has changed (e.g., new product offerings and seasonal patterns). Which concept best explains this issue and what is the most appropriate mitigation?