50 IBM A1000-118 Practice Questions: Question Bank 2025

A project team is explaining AI to non-technical stakeholders. Which statement best describes the difference between AI and machine learning (ML)?

A business analyst asks what a classification model produces. Which output best matches a classification task?

A team wants to build an application that answers users’ questions about internal policy documents using natural language. Which IBM capability is most directly aligned with this need?

A company wants to adopt AI responsibly. Which action best supports transparency for end users impacted by an AI decision?

A dataset for churn prediction has 95% non-churn and 5% churn. The model achieves 95% accuracy but fails to identify most churners. Which metric is most appropriate to focus on to better evaluate performance on the churn class?

A data scientist reports that the training accuracy is very high, but validation accuracy is much lower. What is the most likely issue and a recommended first response?

A team is building an AI feature for customer support. They need a repeatable way to move from data preparation to training and evaluation so results can be reproduced by others. Which practice best supports this goal?

A retail company wants to use AI to personalize product recommendations. Business leaders ask how to ensure the system does not unfairly disadvantage certain customer groups. Which approach best addresses this concern?

A bank is developing a loan-approval model. During testing, the team finds that a feature representing an applicant’s postal code strongly improves accuracy but may act as a proxy for protected characteristics. What is the best next step?

A team built a Watson-based chatbot for IT support. In production, users report that the bot frequently returns confident but incorrect answers, especially for newly updated procedures. Which remediation best aligns with robust AI system design?

A team is starting an AI initiative and wants to clarify the difference between AI, machine learning, and deep learning for stakeholders. Which statement is MOST accurate?

A retailer wants to reduce the manual effort of routing customer emails to the correct department (billing, shipping, returns). Which problem type best matches this use case?

A business user wants to build a simple chatbot for common HR questions without writing code. Which IBM capability is most aligned with this goal?

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

A bank is building a loan approval model and wants to reduce the risk of biased outcomes across demographic groups. Which practice best supports AI fairness?

A team uses Watson Discovery to search across thousands of policy documents. Users report irrelevant results because queries match common words (e.g., “the”, “and”). Which approach is most appropriate to improve relevance?

A product team must choose a success metric for a fraud detection model where only 1% of transactions are fraudulent. Which metric is typically most informative for this imbalanced classification problem?

A company wants an AI system that can justify its recommendations to comply with internal audit requirements. Which model choice and practice best supports this need?

A team trained a sentiment analysis model using customer reviews from one region. After launching globally, performance drops significantly. Which concept best explains this and what is the best first response?

An enterprise wants to deploy an AI solution with strong governance: versioned models, traceable data sources, reproducible training runs, and controlled promotion from development to production. Which approach is most appropriate?

A team is building a classifier and notices that only 5% of records belong to the positive class. Accuracy is 95% even when the model predicts all negatives. Which evaluation metric is most appropriate to understand performance on the minority class?

A business stakeholder asks why the AI team is spending time on data labeling and data quality checks instead of immediately training models. Which response best reflects an AI best practice?

A customer service manager wants an assistant that answers questions using existing policy documents and must cite sources for each answer. Which approach best fits this requirement?

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

An organization wants to ensure its AI solution is used responsibly by defining constraints on how outputs can be applied and by documenting intended use, limitations, and known risks. What is the best artifact to create?

A chatbot for HR frequently gives confident but incorrect answers that are not present in the internal HR handbook. The team wants a practical mitigation without retraining the base model. What is the best next step?

A data science team needs a repeatable process to compare multiple models fairly and avoid using test data during iterative tuning. Which approach is best practice?

A bank plans to deploy an AI model for loan approvals. Regulators require the bank to show that decisions are explainable and that protected groups are not treated unfairly. Which combination best addresses both requirements?

A team trained a model on last year's transaction patterns. In production, fraud tactics change and model performance gradually declines. What is this problem called, and what is an appropriate response?

A company wants to deploy a generative AI assistant that must prevent employees from accidentally exposing sensitive customer data. The assistant will be used across departments with different access rights. Which design choice best supports this requirement?

A customer support team wants to automatically route incoming emails to the correct department (Billing, Technical Support, Sales). They have historical emails labeled by department. Which approach best fits this requirement?

A product manager asks why a model's accuracy decreased after deploying it to production, even though it performed well in testing. Which concept most directly explains this?

A team is building an AI solution and needs to clearly separate training activities from inference in production for stability and governance. Which architecture decision best supports this?

A team wants to build a chatbot that answers employees' HR policy questions using a curated knowledge base of company documents. Which IBM Watson capability is most aligned with this use case?

A bank uses an AI model to recommend credit limits. Regulators require the bank to provide reasons for adverse decisions and to detect potential bias across demographic groups. Which combination best addresses these needs?

During evaluation of a binary classifier for fraud detection, the dataset is highly imbalanced (fraud is rare). Which metric is generally more informative than accuracy for this scenario?

A team is troubleshooting inconsistent model outputs between training and production. Training used one-hot encoding for a 'country' feature, but production uses integer labels (e.g., US=1, CA=2). What is the most likely issue?

A retail company wants to reduce operational risk by ensuring that any AI model deployed can be monitored for drift and performance degradation, with alerts to stakeholders. Which IBM capability is designed for this purpose?

A healthcare organization wants to use patient notes to predict readmission risk. The notes may include protected health information (PHI). Which action is the best first step to support ethical and compliant AI development?

An enterprise wants to adopt a generative AI assistant for internal use. They are concerned about the model producing fabricated but plausible answers. What is the most effective design approach to reduce this risk while keeping responses helpful?

A product team is confused about when to use AI versus traditional software rules. They have a task to route customer emails into 5 categories based on wording and intent, and they expect the wording to change over time. Which approach is most appropriate?

A data scientist notices a model performs extremely well on the training set but significantly worse on new customer data. Which issue is MOST likely?

A business sponsor asks for a simple way to communicate model performance to non-technical stakeholders for a binary classification use case. Which metric is generally the MOST intuitive to start with (assuming classes are reasonably balanced)?

A team is designing an AI assistant that answers questions using internal policy documents. They want responses grounded in those documents and want to reduce hallucinations. Which design pattern best addresses this requirement?

A company wants to implement governance for multiple AI models across teams, including tracking model versions, monitoring quality, and capturing approvals for deployment. Which capability is MOST aligned with this need?

A retail model predicts customer churn. After deployment, the data team discovers the definition of "churn" changed (customers are now considered churned after 60 days of inactivity instead of 30). What is the BEST next step?

A team is evaluating a binary classifier for fraud detection. Fraud is rare (1% of transactions). They report 99% accuracy. Which additional metric is MOST important to review to understand whether the model is actually useful?

A customer wants to deploy an AI solution that must keep personally identifiable information (PII) protected while still enabling model training and analytics. Which practice BEST supports this goal?

A bank deploys a credit decision model and later discovers it rejects a protected demographic group at a significantly higher rate, even though protected attributes were not explicitly included as features. What is the MOST likely explanation?

A company builds a generative AI assistant for employees. A red-team test shows users can craft prompts that cause the model to reveal sensitive content from retrieved internal documents not intended for them. What is the BEST architectural mitigation?