IBM A1000-047 - Assessment: Foundations of AI Practice Exam 2025: Latest Questions

A project team is deciding whether their problem is best solved with traditional programming or AI. Which description best fits an AI approach?

A call center wants to add a virtual assistant that can understand customer questions phrased in many different ways and route them to the right department. Which IBM Watson capability best matches this need?

A data scientist builds a model that predicts whether an email is spam or not spam based on labeled historical examples. What type of machine learning problem is this?

A healthcare organization wants its AI system to provide reasons that clinicians can review before accepting a prediction. Which AI principle is being emphasized?

A team trains a model with 98% accuracy on training data but only 72% on unseen test data. What is the most likely issue?

A company wants to track an AI model in production for drift, bias, and performance over time, and generate explanations for individual predictions. Which IBM capability is best suited for this?

A product manager asks why a model’s overall accuracy is not a good metric for a fraud detection system where fraud is rare. What is the best explanation?

An organization is defining controls to ensure AI systems are used responsibly. Which action is the best example of AI governance?

A team builds a sentiment model and reports strong performance. Later, it’s discovered that during training the dataset contained a feature indicating whether the text came from a 'complaint' form or a 'praise' form—information that would not be available at prediction time. What is the primary problem?

A bank wants to deploy an AI model for loan approvals. Regulators require the bank to demonstrate that protected attributes (such as race) are not directly used and that outcomes are periodically evaluated for fairness. Which approach best addresses this requirement end-to-end?

A team wants an AI system that can explain which input features most influenced each individual prediction for a loan approval model. Which capability best addresses this need?

An online retailer wants to classify customer support tickets into categories such as "billing", "returns", and "shipping" using an IBM-managed service with minimal ML coding. Which IBM capability is the best fit?

A project lead says, "We will measure our model's performance on the same data we used to train it." What is the primary risk with this approach?

A healthcare provider wants to deploy a chatbot that answers common questions but must avoid hallucinated medical advice. Which design choice best reduces the risk while still providing helpful answers?

A data scientist finds that a model performs well overall, but its false negative rate is much higher for one demographic group than another. What should be done first?

A team uses IBM Watson Assistant for a customer service virtual agent. Users report that the assistant often asks unnecessary follow-up questions even when the user already provided the needed detail. Which configuration change is most likely to help?

A model trained on last year's customer behavior performs worse after a major product change. Which concept best explains the issue?

A company wants to extract key fields (e.g., invoice number, total amount, vendor) from many scanned invoice PDFs. Which approach is most appropriate?

A regulated financial institution must ensure its AI model decisions can be audited months later. Which practice best supports this requirement?

A team wants to build a solution that answers questions using internal policy documents and must minimize the chance of exposing confidential passages to unauthorized users. Which architecture choice best addresses this?

A retail team is building an AI model to predict which customers will respond to a promotion. They have only 2% positive responses in historical data. After training, the model reports 98% accuracy but performs poorly in identifying responders. What is the most likely issue and best next step?

A bank wants to provide customers with a chat experience that can answer questions such as "How do I reset my password?" and "What documents do I need for a loan?" The bank requires a solution that supports intent recognition, entity extraction, and integration into channels like web and messaging apps. Which IBM capability best fits this requirement?

A data science team trained a model that recommends job candidates. An audit reveals the model disproportionately filters out applicants from a protected group due to historical bias in the training data. Which action is the best FIRST step aligned with responsible AI practices?

A team is deploying a machine learning model to production. The model’s prediction quality is expected to change over time as customer behavior evolves. Which operational capability is MOST important to detect and address this risk?

A healthcare provider wants to summarize long clinical notes and answer questions grounded ONLY in the patient’s record. They are concerned about hallucinations and need traceability to source text. Which architecture is MOST appropriate?