IBM A1000-076 - Assessment: Foundations of AI Practice Exam 2025: Latest Questions

A retail team wants a clear definition to communicate internally. Which statement best distinguishes Artificial Intelligence (AI) from Machine Learning (ML)?

A chatbot is deployed to answer common HR questions. Users ask, "Can I carry over vacation days?" Which capability is primarily responsible for extracting meaning from the user’s sentence so the system can respond appropriately?

A team is building an image classifier that assigns one label (e.g., "cat" or "dog") to each photo. Which type of machine learning problem is this?

A project sponsor wants an AI solution where predictions can be explained to auditors with simple, human-readable rules. Which model choice best aligns with this requirement?

A bank has a labeled dataset where fraud cases are only 1% of transactions. A model achieves 99% accuracy by predicting "not fraud" for every transaction. What is the most appropriate next step to evaluate the model meaningfully?

A team trains a model and sees low training error but high validation error. Which issue is most likely occurring?

A customer support team wants to automatically route incoming emails to the correct department (Billing, Technical Support, Sales). They have thousands of previously routed emails. Which approach is most appropriate?

A developer wants to build an application that answers questions using passages from a company policy document, and the developer wants the system to cite the relevant text. Which IBM Watson capability best fits this need?

A hiring model shows higher rejection rates for qualified candidates from a protected group. The team wants to reduce discriminatory outcomes while keeping the model useful. Which action is the best first step aligned to responsible AI practices?

An enterprise wants to operationalize an AI solution using IBM services. Requirements: (1) store and manage multiple model versions, (2) deploy models to endpoints, and (3) monitor deployed models for drift and quality over time. Which IBM capability most directly supports these needs?

A product team wants a quick way to explain the difference between AI, machine learning (ML), and deep learning (DL) to non-technical stakeholders. Which statement is MOST accurate?

A bank wants to automatically categorize incoming customer emails into topics such as "card issue," "mortgage," and "fraud." They have historical emails labeled by topic. Which learning approach best fits this problem?

A team is using IBM Watson Assistant for a customer support chatbot. They want the assistant to hand off to a human agent when the user asks for help and the chatbot is not confident. Which feature best supports this requirement?

A data scientist is evaluating a model for detecting rare fraudulent transactions (1% fraud rate). Accuracy is very high, but many fraud cases are missed. Which metric is MOST appropriate to prioritize to reduce missed fraud cases?

A retail company wants to reduce bias in an AI model used to recommend credit limits. Which action is the BEST example of a fairness-focused mitigation step during model development?

A team wants to build an AI solution that extracts key fields (invoice number, date, total) from a variety of invoice PDFs. Which approach is MOST suitable?

A team is using IBM Watson Natural Language Understanding (NLU) to analyze product reviews. They need a single API call that returns sentiment and also identifies entities like product names and brands. What should they configure in their request?

A machine learning model performs well during training but poorly on new, unseen data. Which issue is MOST likely occurring, and what is a common remedy?

A company deploys a model that predicts equipment failure. After several months, the model’s performance steadily degrades because the equipment sensors were recalibrated and the data distribution changed. What is the MOST appropriate practice to address this?

A healthcare organization wants to use a third-party foundation model via an API for summarizing clinical notes. They must reduce the risk of exposing sensitive patient data and meet governance expectations. Which approach is MOST appropriate?

A bank is building an AI system to assist call-center agents by suggesting responses in real time. The bank is concerned that the model might "hallucinate" incorrect policy details. Which best practice most directly reduces this risk while keeping the system helpful?

A retailer wants to group customers into segments based on purchase patterns, without any existing labels. Which machine learning approach is most appropriate?

An HR team trains a model to screen job applicants and notices the model consistently scores candidates from one demographic group lower, despite similar qualifications. What is the most appropriate first step to address this issue?

A team is using IBM Watson Natural Language Understanding to extract insights from customer emails. They only need sentiment and key phrases, and want to minimize implementation effort. What is the recommended approach?

A classification model achieves 95% accuracy on a dataset where 95% of the examples belong to class "A" and 5% to class "B". However, it almost never correctly identifies class "B". Which metric would best highlight this problem?