ISACA AI Fundamentals Practice Test

These questions are part of the practice quiz. Start practicing

• Used for image recognition tasks?
• Which technique increases training data diversity through small automatic transformations?
• List common evaluation metrics for regression tasks.
• Which architecture is especially suitable for modeling sequential data and maintaining context across steps?
• Which term accounts for competitive advantages, long-term strategic value, risk reduction, and financial gains in AI initiatives?
• Which learning paradigm is used to detect novel patterns in data without relying on labeled examples?
• Which statement correctly describes the relationship between KPIs and KGIs?
• What defines data leakage in AI model evaluation?
• Phase where the appropriate algorithm is chosen for the task?
• What is the relationship among artificial intelligence AI, machine learning ML, and deep learning DL?
• Which group provides domain knowledge to ensure data and modeling tasks align with real-world requirements?
• Which metric is not typically used to evaluate classification models?
• Which practice helps prevent data leakage during model evaluation?
• Which subfield uses symbols and explicit rules to represent knowledge for reasoning?
• Biased training data has detrimental social effects. This is described as what?
• What term describes acquiring and refining data according to project stakeholder specifications?
• Which practice focuses on sustainable computing, bias prevention, and model explainability in AI?
• Which regulation is known for China’s comprehensive data protection law?
• What are measurements used for predicting future performance or results?
• Which concept notes that global privacy regulations vary significantly by region?
• Data minimization supports AI governance by which of the following?
• What does few-shot learning entail in AI systems?
• Which privacy-preserving ML technique adds noise to protect individual data?
• What are the parameters within nodes that allow extrapolation in the hidden layers?
• Reacting to unforeseen events outside of the training environment describes which challenge?
• Which regulation right to know what personal data is collected and to delete it?
• Which visual tool displays the counts of correct and incorrect predictions across categories?
• What is batch normalization and its benefit?
• Which concerns relate to the handling and protection of personal data in AI applications?
• What is a practical benefit of model explainability in AI projects?
• What term describes the arrangement where each node processes inputs to produce outputs?
• Which option is NOT a typical threat to AI systems?
• Which metric is appropriate for evaluating a regression model's predictive accuracy?
• Neural networks can be trained on which types of datasets?
• What term describes the overall field of machines performing tasks that typically require human intelligence?
• Which statement about AI sustainability and lifecycle considerations is accurate?
• In the context of image generation, which model is trained through adversarial competition between two networks?
• Which design aims to optimize an AI’s behavior based on feedback from the environment?
• Data drift is best described as a change in data distribution over time affecting model performance.
• Which regulation prohibits apps that pose unacceptable risks like real-time biometric identification?
• Which term refers to the information with which the learning algorithm populates the model during training?
• AI systems can be targets of malicious actors. This risk is known as what?
• What type of ML model is designed to be trained on labeled or unlabeled datasets?
• Which of the following is a defense against AI threats?
• Which component tries to distinguish between the real data in the training set and the generated data?
• Which term denotes a strategic approach to AI development that promotes fairness and reduces bias?
• Which neural network type is designed to highlight patterns in input data through convolutions?
• What is the primary purpose of an AI governance policy?
• NOT a perspective of the Balanced Scorecard?
• Which hardware is vital to the development and deployment of AI systems?
• Which model is trained to assign a class to a sample based on the sample's features?
• If a model has high recall but low precision, which metric best captures the trade-off between them?
• Which approach processes information about similar scenarios and elaborates a new procedure to execute the task?
• Which system type is designed to imitate expert decision processes within a domain?
• Which metric measures the model's ability to detect positive instances among all actual positives?
• Which option best describes a common drawback of machine learning approaches?
• Which is NOT a core principle of Responsible AI?
• Which regulation right to access, correct, erase data, and object to processing in the EU?
• What is the purpose of an AI ethics board?
• Which term describes AI systems that combine rule-based and ML algorithms to achieve accuracy and reliability?
• Which framework is described as optimizing resources, achieving benefits, and reducing risks with a structured approach to AI governance?
• What process creates new sets of features to aid the training model in performing its task?
• Which concept relates to shaping statistical language models and advances in natural language processing?
• What term refers to qualified personnel in software development, statistics, business, and ethics?
• What is data governance and why is it important in AI projects?
• In AI development, what is the difference between validation and verification?
• The emphasis on sustainable computing, bias prevention, and model explainability corresponds to which practice?
• Which term denotes the collection of values chosen before training that govern the learning process?
• Which of the following is NOT a common bias type in AI?
• Which learning approach works by adjusting behavior to labels and features in a test set generated automatically?
• Which option provides the essential foundation for the development and deployment of AI models?
• Which regulation right to access, correct, delete, and transfer data?
• Which term describes retrospective metrics that measure past performance outcomes?
• Which concept is a tactical approach to developing and using AI tools ensuring diversity and reducing bias?
• In a Generative Adversarial Network, which component is responsible for producing counterfeit data to challenge the discriminator?
• Which learning approach uses both labeled and unlabeled data to improve model performance?
• Which technique increases the size and diversity of the training data through small automatic transformations?
• Which metric is the harmonic mean of precision and recall?
• Which concept focuses on privacy, bias, responsibility, and the nature of intelligence and autonomy in AI?
• How do ISO and NIST AI standards influence AI governance?
• What term describes the learning algorithm's iteration through all samples in the dataset?
• Less predictable, require much data and computational resources, post greater threat to environment
• Which statement best describes a neural network?
• Which statement best describes data augmentation?
• Model versioning involves what?
• Which neural networks process input data through convolutions to highlight relevant patterns?
• Which statement best distinguishes interpretability from explainability in AI?
• Which analysis is used to justify AI investments by projecting competitive advantages and financial gains?
• What term describes toolkits for deep learning that offer pre-built modules and optimization algorithms?
• Which role ensures that data accurately reflects the task for which the model is trained?
• Which metric measures overall correctness as correct predictions divided by total samples?
• Which lifecycle stage involves decommissioning models and data?
• What measure of machine intelligence was proposed in the book that introduced the Turing Test?
• Which metric avoids false negatives, calculated as correct positive predictions divided by total actual positives?
• Which architecture is typically favored for tasks involving sequential data and memory of past elements?
• Which concept refers to AI capable of performing any intellectual task that a human being can do?
• Phase where the model's performance is evaluated?
• Measurements that show how well a process is working toward a specific objective are known as what?
• Which statement best describes privacy-preserving inference?
• What are the small processing components in a neural network called?
• Which activity demonstrates static analysis in ML systems?
• Data provenance is essential because it provides:
• What is model calibration and what problem does it address?
• Name common data privacy techniques used in machine learning.
• Which term describes governance focusing on privacy, bias, responsibility, and autonomy in AI?
• Which algorithm is used to predict a numeric value from input features?
• Which architecture can process many input features in parallel?
• List common evaluation metrics for classification tasks.
• What is a decision boundary in classification?
• What ensemble model combines multiple decision trees to improve accuracy?
• What is online learning versus batch learning?
• Which technique is used for highly realistic image generation?
• Which process involves forward pass, loss function, backpropagation, optimization, and continues until the loss value is negligible?
• Which component tries to generate fake data that the discriminator cannot distinguish from real data?
• What is Explainable AI and why is it important for governance?
• Which technique trains models locally on devices and aggregates updates?
• What is a risk register in AI governance?
• Which prompting approach involves providing a description of the thought process used by the model?
• What is overfitting and how can it be mitigated?
• Which technique is foundational for training large language models through predicting masked tokens?
• The framework that includes customer, financial, internal process, and learning and growth perspectives is known as what?
• What term describes models that capture and automate expert procedures within a domain?
• Which statement correctly differentiates disparate impact and discriminatory outcome?
• Which process requires specialized hardware, large infrastructure, and a constant supply of energy?
• What term refers to AI that is designed to operate within a single domain or task area?
• Which term is used for AI that aims to replicate full human-level intelligence across diverse tasks?
• Which statement correctly contrasts leading indicators and lagging indicators?
• What are adversarial examples in AI security?
• Which subset of machine learning involves neural networks with many layers?
• Which method uses probabilities to reason under uncertainty about the real world?
• Which technique adds noise to data to protect privacy?
• Which term best describes a field that integrates rule-based systems with data-driven learning to improve reliability?
• Which statement best describes AI sustainability?
• What is the process called that involves adjusting the network's parameters during learning?
• What issue arises when training data is biased, potentially amplifying social inequities?
• Which term refers to preparing data for modeling by addressing missing values and inconsistencies?
• How can data drift affect model performance over time?
• Which concept is defined as balancing technological proficiency with ethical considerations, legal compliance, and strategic alignment?
• Which role is responsible for risk/compliance function in AI governance?
• Which approach focuses on applying probabilistic methods to solve complex problems?
• Which practice supports auditability in AI systems?
• What is regularization and name two examples.
• Which type of neural networks is designed to handle sequential data more effectively than traditional neural networks?
• Which term describes the iterative procedure that updates model parameters by computing gradients and minimizing loss?
• Which elements constitute a monitoring plan for AI systems?
• Which of the following are common variants of gradient descent?
• In AI security testing, what are red team and blue team roles?
• Which of the following metrics are derived from a confusion matrix?
• What is gradient descent?
• Which statement correctly distinguishes static analysis from dynamic analysis in ML systems?
• First phase in ML project to frame the issue being addressed?
• Which of the following is NOT listed as a major stage in the AI lifecycle?
• Which option best describes a key advantage of rule-based systems?
• Which phase in the ML project lifecycle corresponds to the model learning from data?
• Which issue arises when the model fits the training data extremely well but does not generalize to new data?
• Which term focuses on establishing data usage policies and aligning data with organizational goals?
• Which approach focuses on learning from data by adjusting to labeled examples?
• Which concept is about establishing clear data usage policies and ensuring data aligns with organizational goals?
• Which set of elements are essential components of an AI governance policy?
• Which technique enables performing computations on encrypted data without decrypting?
• Which privacy technique adds controlled random noise to protect individuals information when sharing model outputs?
• Which issue occurs when the learning algorithm fails to extrapolate a functional set of parameters from the training data?
• Which architecture is used for natural language processing?
• What is human-in-the-loop oversight and why is it important?
• Which statement best describes supervised learning?
• Which model is designed for classification and can also be applied to regression problems?
• Which metric avoids false positives, calculated as correct positive predictions divided by total positive predictions?
• What term describes written or spoken requests entered into a generative AI system to obtain a response?
• Which statement correctly distinguishes data drift from model drift?
• Unintended or unauthorized deletion, corruption, alteration, or loss of data is called what?
• What is dropout in neural networks?
• Which term is used for the basic computational unit in a neural network that processes inputs and emits an output?
• Which technique keeps user data on local devices while contributing to a global model?
• What is transfer learning and when is it useful?
• Which approach is grounded in representing knowledge with symbols and using logical inference?
• Which technique is widely used in the development of large language models?
• Which practice helps detect model drift over time?
• Which AI learning method is designed to replicate behaviors, inferences, or decisions demonstrated by a collection of samples?
• Which artifact provides the counts of predictions versus actual outcomes to facilitate error analysis?
• Which hardware are specialized circuits explicitly designed to accelerate deep learning workloads?
• The description 'Brainstorming sessions, concept ideation, customized media' is associated with which concept?
• Which term refers to the set of parameters that the algorithm learns from data during training?
• What is data augmentation and its purpose in ML?
• Which description best supports data lineage’s role in trust and auditability?
• Which architecture uses attention mechanisms to enable parallel processing of input features?
• Which technique adds random noise to data to protect privacy?
• Which term refers to the layers between input and output that enable learning?
• Which branch of AI has the capability of generating new and original content?
• Which process involves removing or replacing missing values, tackling outliers, and eliminating errors or inconsistencies?
• Which framework is described as optimizing resources, achieving benefits, and reducing risks with a structured approach to AI governance?
• What does the bias-variance trade-off describe in model development?
• Which technique trains AI models across devices without sharing raw data?
• What term describes the settings that determine how the learning algorithm processes data to populate the model?
• Which ensemble method uses multiple decision trees and reduces variance by averaging?
• What term describes media that show real or imagined individuals, often used for fake news?
• In AI governance, which term describes ongoing oversight after deployment?
• Which of the following is NOT a privacy-preserving ML technique listed?
• What consists of two AI models, a generator and a discriminator?
• Assessing benefits and requirements for AI implementation is best described by which term?
• Which phase in ML project is explicitly responsible for selecting the algorithm to use?
• Highly dependable and predictable, easily interpretable and modifiable, inexpensive
• Which is typically the first stage in the AI lifecycle?
• Which component specifically enables explanations of AI decision processes to users?
• What circuits are optimized to run deep learning algorithms?
• What is a red team exercise in AI risk assessment?
• What type of recurrent neural network is used in machine learning?
• What is cross validation used for in model evaluation?
• Which description reflects the characteristics of effective KPIs?
• Which statement best describes the purpose of KPIs in AI initiatives?
• What is the purpose of governance compliance reviews in AI monitoring?
• What are the layers between input and output that are not directly visible called?
• Which paradigm learns by trial and error through reward signals to improve its policy?
• What is a secure enclave?
• What does trustworthy AI entail?
• What is dataset leakage and how can it be prevented?
• Neural networks with multiple hidden layers are known as?
• What functions let the network reduce the impact of a feature or combination of features?
• Which method trains AI models across devices without sharing raw data?
• Which algorithm is commonly used for classification and can operate with margin-based optimization?
• Which concept notes a balance between technological proficiency, ethical considerations, legal compliance, and strategic alignment?
• If a metric specifies the outcomes to be achieved, it is best described as a KGI rather than a KPI. Which option reflects this idea?