Free CAIC Practice Test Questions and Answers (2026)

Natural Language Processing (NLP) encompasses a wide spectrum of tasks, from foundational preprocessing steps to complex applications. Tokenization (breaking text into units) and Part-of-Speech (PoS) tagging (assigning grammatical labels) are fundamental sub-tasks essential for many NLP pipelines. Question Answering (QA) is a higher-level, application-oriented task that aims to answer user queries in natural language. All three—Tokenization, PoS tagging, and QA—are well-established and correct examples of tasks within the domain of NLP. Therefore, the option including all three is the most complete and accurate answer.

A. This is incomplete. While tokenization is an NLP task, so are PoS tagging and QA, making this option not the best choice.

B. This is incomplete. PoS tagging is a correct NLP task, but the other options are also valid tasks within the NLP domain.

C. This is incomplete. QA is a correct NLP task, but it omits the foundational tasks also listed as options.

D. This is incomplete as it correctly identifies two tasks but omits the valid high-level task of Question Answering (QA).

1. Jurafsky, D., & Martin, J. H. (2023). Speech and Language Processing (3rd ed. draft). Chapter 2 discusses tokenization, Chapter 8 covers Part-of-Speech tagging, and Chapter 23 details Question Answering.

2. Manning, C. D. (2022). CS224N: Natural Language Processing with Deep Learning, Lecture 2: Word Vectors and Word Senses. Stanford University. This lecture introduces foundational NLP tasks, including tokenization and tagging, as prerequisites for more advanced applications.

Few-shot learning is a machine learning approach where a model must make accurate predictions for a new task despite having seen only a very small number of labeled examples. This necessitates two key conditions: the number of examples is small, typically just a handful (B), and the model must leverage a strong base of prior knowledge, often gained through pre-training on a large, diverse dataset, to generalize effectively from these few examples (C). Both statements are essential to correctly define the concept.

A. This statement is incorrect because "a large number of examples" is the opposite of the "few-shot" paradigm.

B. This statement is correct but incomplete on its own, as it doesn't explain how the model accomplishes the task with so few examples.

C. This statement is also correct but incomplete, as it describes the mechanism without defining the "few-shot" condition of having very limited examples.

D. This option is incorrect because statement A is fundamentally wrong.

1. Wang, Y., Yao, Q., Kwok, J. T., & Ni, L. M. (2020). Generalizing from a few examples: A survey on few-shot learning. ACM Computing Surveys (CSUR), 53(3), 1-34. (Section 1, "Introduction," defines few-shot learning as learning from a few labeled instances and emphasizes the role of prior knowledge). DOI: https://doi.org/10.1145/3386252

2. Brown, T. B., Mann, B., Ryder, N., et al. (2020). Language Models are Few-Shot Learners. arXiv preprint arXiv:2005.14165. (The abstract and introduction describe the ability of large models to perform tasks given only a few examples, highlighting both the small sample size and the reliance on knowledge learned during pre-training).

Statements A and B are incorrect because they mischaracterize the primary function of retrospective/concurrent risk methodologies. These methods are focused on assessment—analyzing historical (retrospective) and current (concurrent) model performance to measure and understand existing risk. They answer the question, "How has the model performed, and how is it performing now?" While this assessment is a critical input for predicting future risk, the act of prediction itself is a separate, prospective function. Prospective methods use the findings from retrospective/concurrent analysis to forecast future performance, often through stress testing or simulation. Therefore, stating that retrospective/concurrent methods predict future risk is functionally incorrect.

A. Retrospective/concurrent risk methods predict model risk...: This is incorrect. These methods assess past and present risk; prediction is a prospective activity that uses this assessment as an input.

B. Retrospective/concurrent risk leverages the most current risk... to predict... for future cycles: This is incorrect for the same reason as A. It conflates the assessment of current risk with the separate function of predicting future risk.

C. Retrospective/concurrent is more suitable when...: This statement is correct. It accurately describes a key use case for these methods—investigating a model's performance after significant changes or incidents.

E. a, b and c only: This is incorrect because statement C is a correct description of when to apply the methodology.

1. National Institute of Standards and Technology. (2023). AI Risk Management Framework (AI RMF 1.0). NIST AI 100-1. https://doi.org/10.6028/NIST.AI.100-1 (See Section 3.3, "MEASURE," which discusses tracking metrics over time, aligning with retrospective/concurrent assessment, while prospective analysis is part of managing future risks).

2. Board of Governors of the Federal Reserve System. (2011). Supervisory Guidance on Model Risk Management. SR 11-7. (This foundational document in model risk management distinguishes between initial validation, ongoing monitoring (concurrent/retrospective), and activities like stress testing (prospective) to understand future behavior. See page 6 on "Ongoing Monitoring").

3. Engle, R. F., & Manganelli, S. (2004). CAViaR: Conditional Autoregressive Value at Risk by Regression Quantiles. Journal of Business & Economic Statistics, 22(4), 367-381. https://doi.org/10.1198/073500104000000370 (This paper exemplifies risk modeling where backtesting is a retrospective analysis to validate a model's past predictions, distinct from the model's function of forecasting future risk).

Business goals are high-level objectives that a company aims to achieve. Artificial intelligence is a strategic tool used to help meet these goals. Reducing operational costs, mitigating business risks (such as fraud or compliance issues), and improving revenue through better products or services are all fundamental and universally recognized business goals. AI can be applied to optimize processes, enhance decision-making, and create new value propositions, directly contributing to each of these objectives. Therefore, all the listed options are correct examples of business goals that can be addressed with AI solutions.

A. This is a valid business goal, but not the only one listed.

B. This is also a valid business goal, but not the only one listed.

C. This is another valid business goal, but not the only one listed.

D. This option is incomplete as it omits the equally valid goal of revenue improvement.

1. Davenport, T. H., & Ronanki, R. (2018). Artificial Intelligence for the Real World. Harvard Business Review, 86(1), 108-116. The article categorizes AI initiatives into process automation (cost reduction), cognitive insight (risk mitigation), and cognitive engagement (revenue improvement).

2. Iansiti, M., & Lakhani, K. R. (2020). Competing in the Age of AI: Strategy and Leadership When Algorithms and Networks Run the World. Harvard Business Review Press. Chapter 2, "Reinventing the Business," discusses how AI transforms core business operations to achieve goals like efficiency (cost), scale, and new revenue streams.

3. MIT Sloan Management Review. (2019). Winning With AI. This report outlines key business objectives for AI adoption, including creating new products/services (revenue), improving existing ones, and optimizing internal operations (cost/risk).

An Industry Architect is a specialized role focused on designing solutions for a specific industry, leveraging deep domain knowledge of that industry's challenges, regulations, and business goals. The statement in option B describes DevOps, which is a set of practices and a cultural philosophy that combines software development (Dev) and IT operations (Ops). DevOps aims to shorten the systems development life cycle and provide continuous delivery with high software quality. This is a process-oriented approach, not a specific architectural role tied to an industry vertical.

A. This is a correct definition. Industry architects are specialists who tailor solutions to specific industry verticals like finance, healthcare, or manufacturing.

C. This correctly describes the core competency of an Industry Architect, which is deep, domain-specific knowledge and expertise.

D. This accurately portrays the collaborative nature of the role, working with various stakeholders to align technology with industry-specific business needs.

E. This correctly states the primary function of an Industry Architect: translating industry-specific business requirements into viable technical architectures.

1. Lenk, A., et al. (2009). The Enterprise Architecture Sourcebook. The Open Group. Section 3.3, "Architecture Roles," discusses specialized architect roles.

2. Humble, J., & Farley, D. (2010). Continuous Delivery: Reliable Software Releases through Build, Test, and Deployment Automation. Addison-Wesley Professional. Chapter 1 provides a foundational definition of the principles that would become known as DevOps.

3. Lwakatare, L. E., et al. (2019). DevOps in practice: A multiple case study of five companies. Information and Software Technology, 114, 217-230. https://doi.org/10.1016/j.infsof.2019.06.010. This paper defines DevOps as a collaborative approach between development and operations.

Semi-supervised learning is a machine learning paradigm that utilizes a combination of a small amount of labeled data with a large amount of unlabeled data for training. The model learns the underlying structure from the unlabeled data and uses the labeled data to ground its predictions, effectively propagating labels to the unlabeled points. This approach is ideal when data labeling is expensive or time-consuming. The question, despite its awkward phrasing ("unlabeling the data"), describes a scenario where a machine learns to label data, which is characteristic of this method when a mix of labeled and unlabeled inputs are available.

A. Supervised learning requires a fully labeled dataset for training, which is not the case here.

B. Unsupervised learning works with entirely unlabeled data to find patterns, not to assign predefined labels.

C. Reinforcement learning involves an agent learning through rewards and penalties, which is a different paradigm.

D. Semi-reinforcement learning is not a standard, widely recognized term for this specific machine learning category.

1. Chapelle, O., Schölkopf, B., & Zien, A. (2006). Semi-Supervised Learning. MIT Press. In the Introduction (Chapter 1, p. 1), the book defines semi-supervised learning as the case where the learner is given a set of labeled examples and a set of unlabeled examples.

2. Zhu, X. (2005). Semi-Supervised Learning Literature Survey. University of Wisconsin-Madison, Computer Sciences Technical Report 1530. Section 1, "Introduction," defines semi-supervised learning as learning from a combination of labeled and unlabeled data.

3. Van Engelen, J. E., & Hoos, H. H. (2020). A survey on semi-supervised learning. Machine Learning, 109(2), 373-440. https://doi.org/10.1007/s10994-019-05855-6. The abstract clearly defines the paradigm as leveraging unlabeled data in addition to labeled data.

A soft launch is a strategic, limited release of a product to a specific audience before a full-scale launch. Statement (A) correctly defines it as a controlled release. Statement (B) accurately describes the target audience as a select, often handpicked group. Statement (C) correctly identifies the primary goals: to gather real-world user feedback, test the product's viability, and measure key performance indicators like Customer Acquisition Cost (CAC) in a low-risk environment. Since all three statements accurately describe the definition, audience, and goals of a soft launch, option E is the most complete answer.

A. This is correct but incomplete; B and C also describe essential aspects of a soft launch.

B. This is correct but incomplete; A and C are also valid statements about a soft launch.

C. This is correct but incomplete; A and B define the "what" and "who" of a soft launch.

D. This is correct but incomplete, as it omits the crucial goals described in statement C.

1. Ries, E. (2011). The Lean Startup: How Today's Entrepreneurs Use Continuous Innovation to Create Radically Successful Businesses. Crown Business. (Chapter 8 discusses validated learning through early product releases to measure real customer data).

2. Stanford University. (n.d.). Lecture 5: Value Proposition [Lecture notes]. CS183C, Technology-enabled Blitzscaling. Retrieved from Stanford's courseware, which discusses iterative product release strategies like soft launches for gathering customer feedback and metrics.

The Well-Architected Framework, originally developed by AWS and widely adopted, is structured around a set of core pillars that represent key principles for designing and operating cloud systems. The standard pillars are Operational Excellence, Security, Reliability, Performance Efficiency, and Cost Optimization. While frameworks are adapted for specific workloads like Generative AI, they build upon these core pillars. "System Architecture Excellence" is the overall goal or outcome of applying the framework's principles; it is not a pillar itself.

A. Operational Excellence is a core pillar, focusing on running and monitoring systems to deliver business value and continuously improving processes.

B. Security is a core pillar. For GenAI, this is often expanded to explicitly include Privacy due to the sensitive nature of training and generated data.

C. Reliability is a core pillar, ensuring a system recovers from failures and meets its availability and consistency targets.

E. This is incorrect because option D is not a pillar of the Well-Architected Framework.

1. Amazon Web Services. (2023). AWS Well-Architected Framework. AWS Whitepaper. This document defines the six pillars: Operational Excellence, Security, Reliability, Performance Efficiency, Cost Optimization, and Sustainability. (Available at: https://docs.aws.amazon.com/wellarchitected/latest/framework/the-pillars-of-the-framework.html)

2. Amazon Web Services. (2023). Generative AI on AWS. AWS Well-Architected Lens. This guidance applies the existing Well-Architected pillars to the specific challenges and considerations of building GenAI applications. It does not introduce new pillars.

3. Microsoft Azure. (2023). Microsoft Azure Well-Architected Framework. Microsoft Learn. This documentation outlines Microsoft's five pillars, which are very similar to the AWS model: Cost Optimization, Operational Excellence, Performance Efficiency, Reliability, and Security.

Workflow automation in AI, often a core part of MLOps (Machine Learning Operations), involves creating a resilient and repeatable system for the entire machine learning lifecycle. This requires designing and managing the pipeline of steps (A), from data ingestion to model deployment. It also necessitates robust mechanisms for executing these pipelines and monitoring their operational health (B). A critical, specific part of this is configuring monitoring for the model itself (C), such as detecting data drift or performance degradation, to ensure its continued reliability and trigger retraining. All three are integral components for a complete automation solution.

A. Pipeline design and management is a key component, but it is not the only one.

B. Pipeline execution and monitoring is also a key component, but it is not the only one.

C. Model monitoring is a crucial part of the overall workflow, but it is not the only component.

D. This option is incomplete as it omits the critical aspect of specific model monitoring configuration.

1. Google Cloud Documentation. (2020). MLOps: Continuous delivery and automation pipelines in machine learning. The section "ML pipeline automation" describes the necessity of designing, executing, and monitoring automated pipelines for CI/CD in machine learning.

2. Kreuzberger, D., et al. (2022). Machine Learning Operations (MLOps): Overview, Definition, and Architecture. arXiv preprint arXiv:2205.02302. https://doi.org/10.48550/arXiv.2205.02302. Section 3, "MLOps Architecture," details components including automated ML pipelines (design, execution) and continuous model monitoring.

3. Microsoft Azure Documentation. What is MLOps?. This official documentation outlines MLOps principles, which include automating the end-to-end ML lifecycle (pipeline design/execution) and monitoring models for operational and performance issues.

The role of a DevOps architect is fundamentally technical, focusing on the infrastructure, processes, and tools that enable efficient and reliable software development and operations. Key responsibilities include designing CI/CD pipelines, implementing monitoring and alerting systems, and planning for disaster recovery. Business development and strategic business planning are outside this technical scope; they are typically handled by roles like product managers, business analysts, or sales executives. While a DevOps architect must understand business goals, they do not own the business development function.

A. Monitoring and alerting are fundamental DevOps practices for maintaining system health and operational awareness, making this a core responsibility.

C. Disaster recovery planning is a critical aspect of ensuring system reliability and resilience, a key concern for a DevOps architect.

D. This statement accurately describes the implementation aspect of disaster recovery, which falls squarely within the DevOps architect's purview.

E. CI/CD pipelines are the backbone of DevOps, automating the software delivery lifecycle, and their design is a primary duty of the architect.

1. Humble, J., & Farley, D. (2010). Continuous Delivery: Reliable Software Releases through Build, Test, and Deployment Automation. Addison-Wesley Professional. Chapter 1, "The Problem of Delivering Software," outlines the core technical challenges that DevOps and roles like the DevOps architect are meant to solve.

2. Kim, G., Behr, K., & Spafford, G. (2013). The Phoenix Project: A Novel About IT, DevOps, and Helping Your Business Win. IT Revolution Press. The book illustrates through narrative the core technical and cultural responsibilities of DevOps roles, focusing on workflow, feedback, and continuous improvement, not business development.

3. University of California, Berkeley. (n.d.). CS 169: Software Engineering. Course materials on modern software development practices cover DevOps principles, emphasizing automation (CI/CD), monitoring, and infrastructure as code, defining the technical scope of related roles.