IBM A1000-108 - Assessment: Foundations of AI and Machine Learning Practice Exam 2025: Latest Questions

A retail team wants to automatically tag product photos with labels like "shoe," "hat," and "bag" using previously labeled images. Which AI approach best fits this requirement?

A data scientist builds a model that achieves 98% accuracy on the training set but only 72% on new validation data. What is the most likely issue?

A team is preparing a dataset for a customer churn model and notices that some rows have missing values in the "MonthlyCharges" field. What is a recommended first step before choosing how to handle the missing values?

Which statement best describes the difference between AI and machine learning?

A bank is building a model to predict whether a transaction is fraudulent (yes/no). Fraud cases are rare (about 1% of transactions). Which evaluation metric is generally more informative than accuracy for this scenario?

A team trains a model using customer data. Later, they discover that "Churn" was accidentally included as an input feature during training. What problem does this introduce?

An insurance provider wants to ensure its AI system for claim triage is transparent to auditors. Which practice most directly supports transparency and explainability?

A model is deployed to predict product demand. After a few months, it starts performing poorly because customer behavior changes due to a new competitor. What is the best operational response?

A hiring screening model shows a significantly lower selection rate for one protected group compared to others. The team wants to reduce discriminatory impact while keeping the model useful. Which approach is most appropriate?

A team built a sentiment classifier for customer reviews. In production, accuracy drops sharply. Investigation shows the training data was mostly formal written reviews, but production data includes slang, emojis, and short phrases. What is the most likely root cause?

A retail team wants to group customers into segments based on purchasing behavior, but they do not have pre-labeled segment names. Which type of machine learning task fits this goal?

A dataset contains many missing values in a numeric feature such as 'annual_income'. Which approach is generally a reasonable first baseline for handling the missing values before model training?

An executive asks, "What is the difference between AI and machine learning?" Which statement is most accurate?

A bank trains a loan-approval model and sees excellent accuracy on the training set but significantly worse performance on new applicants. Which issue is most likely occurring?

A team is building a chatbot. They notice that the model frequently generates plausible-sounding but incorrect answers. Which evaluation approach best helps quantify this type of issue for a question-answering use case?

A dataset includes a categorical feature 'country' with hundreds of possible values. For a baseline linear model, which encoding approach is most appropriate?

A company is deploying an AI system to assist with hiring decisions. Which action best aligns with responsible AI practices to reduce risk of unfair outcomes?

A data scientist splits time-series data randomly into training and test sets and reports strong test performance. Later, the model performs poorly in production. What is the most likely cause?

A healthcare organization wants to train a diagnostic model while minimizing exposure of sensitive patient data. Which approach best supports this goal while still enabling model training across multiple sites?

A model is deployed to predict equipment failures. After several months, technicians report that predictions are less accurate because operating conditions have changed (new suppliers, new usage patterns). What is the best practice response?

A retail team is building a supervised model to predict whether a customer will churn. The dataset contains 5% churners and 95% non-churners. The model achieves 95% accuracy but rarely identifies churners. Which evaluation approach is the BEST next step?

A team is preparing data for a model that predicts home prices. The dataset includes a feature called "neighborhood" with values like Downtown, Suburban, and Rural. What is the MOST appropriate way to represent this feature for a typical linear regression model?

A bank wants to deploy an AI system to help decide loan approvals. Regulators require that the bank can provide understandable reasons for decisions to individual applicants. Which model choice BEST aligns with this requirement?

A data scientist reports that a classification model performs extremely well on the training set but significantly worse on the test set. Which issue is MOST likely occurring, and what is a best-practice mitigation?

A team trains a model to predict whether patients are at risk of readmission. They accidentally compute feature scaling parameters (mean and standard deviation) using the full dataset before splitting into train/test. The test performance looks unusually high. What is the BEST explanation and fix?