Free IBM AI Engineering practice test

A. × Incorrect: This statement contradicts the importance of assessing risks in engineering decisions related to AI.

B. × Incorrect: Avoiding review goes against the practice of ensuring responsible and ethical use of AI systems.

C. ✓ Correct: It accurately reflects IBM's guidelines on evaluating AI for responsible use, fairness, explainability, and risk management.

D. × Incorrect: This statement dismisses critical aspects of ethical AI development.

A. × Incorrect: Avoiding all review contradicts the principles of responsible AI development.

B. × Incorrect: Ignoring fairness and explainability goes against best practices in ethical AI design.

C. ✓ Correct: It accurately represents the criteria for evaluating AI systems according to IBM's guidelines.

D. × Incorrect: Engineering decisions are crucial in determining the risks associated with AI.

A. × Incorrect: Supervised learning requires fully labeled data for training models.

B. × Incorrect: Unsupervised learning does not use any labels during training and focuses on finding hidden patterns without guidance.

C. ✓ Correct: Self-supervised learning uses partially labeled or unlabeled data to learn from the structure of input data itself.

D. × Incorrect: Reinforcement learning involves an agent interacting with an environment and learning through trial and error.

A. ✓ Correct: Recurrent Neural Networks (RNNs) have memory capabilities that allow them to maintain state information over time, making them suitable for tasks involving sequential data.

B. × Incorrect: Convolutional layers are used in convolutional neural networks for detecting local patterns and not specifically designed for handling sequential data.

C. × Incorrect: Pooling layers reduce the spatial dimensions of feature maps and do not contribute to memory capabilities or sequence prediction tasks.

D. × Incorrect: Feature extraction involves transforming raw input into meaningful features but does not inherently involve memory or sequence processing.

A. ✓ Correct: It directly addresses the key considerations needed to ensure responsible use of an AI system.

B. × Incorrect: It suggests a separation between AI risk and engineering decisions, which contradicts ethical practices in AI development.

C. × Incorrect: Avoiding all review goes against the principles of responsible AI, which requires continuous evaluation and improvement.

D. × Incorrect: Fairness and explainability are crucial aspects of ensuring that an AI system operates ethically and transparently.

A. ✓ Correct: Precision measures the proportion of true positive predictions out of all predicted positives.

B. × Incorrect: Recall measures the fraction of correctly identified positives among all actual positives, not predicted ones.

C. × Incorrect: Accuracy assesses overall correctness but does not specifically measure positive prediction accuracy like Precision does.

D. × Incorrect: F1 Score balances precision and recall but does not solely focus on the proportion of true positives as Precision does.

A. × Incorrect: Supervised learning uses labeled data to train models, not neural network architectures.

B. × Incorrect: Unsupervised learning finds hidden patterns without using labels but does not specify neural networks.

C. × Incorrect: Reinforcement learning involves decision-making through trial and error, not pattern recognition with neural networks.

D. ✓ Correct: Deep Learning uses neural network architectures to learn complex patterns from data.

A. ✓ Correct: Convolutional Neural Networks (CNN) are specifically designed to handle spatial hierarchies in visual data, which makes them suitable for image recognition tasks.

B. × Incorrect: Recurrent Neural Networks (RNNs) specialize in sequence prediction tasks involving sequential data like time series analysis or natural language processing.

C. × Incorrect: Support Vector Machines (SVMs) are primarily used for classification and regression problems, not pattern detection in visual data.

D. × Incorrect: Decision Trees are typically used for decision-making based on a set of rules derived from the data, rather than recognizing patterns in images.

A. × Incorrect: Ensuring data leakage does not occur is important but does not directly address fairness in hiring processes.

B. × Incorrect: Implementing decision trees without pruning can lead to overfitting and does not ensure fairness in classification models.

C. ✓ Correct: Using synthetic data to balance class representation helps mitigate bias by ensuring a more balanced dataset, which is crucial for fairness in AI systems used in hiring.

D. × Incorrect: Applying gradient boosting for better accuracy focuses on model performance rather than fairness considerations.

A. ✓ Correct: Supervised learning involves training models with labeled data, such as historical sales data, to predict future trends accurately.

B. × Incorrect: Unsupervised learning does not use labels or predefined outcomes; it focuses on finding hidden patterns in data without using labels.

C. × Incorrect: Semi-supervised learning uses a combination of labeled and unlabeled data but primarily relies on the limited amount of labeled data available, which is not specified here.

D. × Incorrect: Reinforcement learning involves training models through trial and error to maximize rewards, rather than predicting trends based on historical data.