Free APMG-International Artificial-Intelligence-Foundation Practice Test Questions and Answers (2026)

Ada Lovelace is universally recognized as the pioneer of computer programming. In the 1840s, she collaborated with Charles Babbage on his proposed mechanical general-purpose computer, the Analytical Engine. Lovelace was the first to recognize that the machine had applications beyond pure calculation, publishing the first algorithm intended to be carried out by such a machine (specifically, to compute Bernoulli numbers). Her visionary insight that numbers could represent entities other than quantity laid the conceptual foundation for modern computing and artificial intelligence. Consequently, she is celebrated in computer science history as the world's first computer programmer, long before electronic computers were invented.

A. Dame Wendy Hall is a prominent contemporary computer scientist known for multimedia and web science, not the historical pioneer of programming.

B. Karen Spärck Jones was a pioneer in information retrieval and natural language processing, not the first computer programmer.

D. Sophie Wilson is a highly influential computer scientist who co-designed the ARM architecture, but she is not the pioneer of programming.

1. Fuegi, J., & Francis, J. (2003). "Lovelace & Babbage and the creation of the 1843 'notes'". IEEE Annals of the History of Computing, 25(4), 16-26. https://doi.org/10.1109/MAHC.2003.1253887

2. APMG International. "Artificial Intelligence (AI) Foundation Syllabus", Section on the History of AI and Computing.

3. Stanford University. (n.d.). "Ada Lovelace", Stanford Computer Science History Courseware (CS101).

The development of statistics in the 18th and 19th centuries heavily featured the formalization of Bayes' theorem, named after Reverend Thomas Bayes and later expanded by Pierre-Simon Laplace. Bayes' theorem is a foundational mathematical formula used extensively in probabilistic inference and modern artificial intelligence. It describes the probability of an event based on prior knowledge of conditions that might be related to the event. In machine learning, Bayesian inference provides a principled way to update beliefs or model parameters as new data becomes available. This theorem forms the backbone of various AI algorithms, including Naive Bayes classifiers and Bayesian networks, enabling systems to handle uncertainty and make reasoned predictions based on statistical evidence.

A. Boltzmann's work relates to statistical mechanics and thermodynamics in physics, not the primary theorem for probabilistic inference in AI.

B. Kolmogorov developed the modern axiomatic foundations of probability theory in the 1930s, not during the 1800s.

D. The Central Limit Theorem describes the distribution of sample means, whereas Bayes' theorem is the specific foundation for probabilistic inference.

1. Russell, S., & Norvig, P. (2020). Artificial Intelligence: A Modern Approach (4th ed.). Pearson. Chapter 13: Quantifying Uncertainty, pp. 412-416.

2. MIT OpenCourseWare, 6.041 Probabilistic Systems Analysis and Applied Probability, Lecture 3: Bayes' Theorem, p. 5.

Diagnostic image analysis is a premier application of Machine Learning (ML) within the healthcare sector. By utilizing deep learning techniques, specifically Convolutional Neural Networks (CNNs), AI systems can analyze complex medical imagery such as X-rays, MRIs, and CT scans with high precision. These ML models are trained on vast datasets of annotated medical images, enabling them to identify patterns, anomalies, and early indicators of diseases like cancer or neurological disorders. This capability significantly augments the diagnostic accuracy of radiologists, reduces human error caused by fatigue, and accelerates the triage process for critical cases, thereby directly improving patient outcomes and operational efficiency in modern healthcare settings.

A. Autonomous wheelchairs: This primarily involves robotics and spatial navigation rather than the core pattern recognition typical of healthcare ML diagnostics.

B. Automated blood sampling: This is a mechanical and robotic automation task, not primarily a Machine Learning application.

C. Autonomous vehicles: While heavily reliant on ML and computer vision, this is an application in the transportation sector, not healthcare.

1. Topol, E. J. (2019). High-performance medicine: the convergence of human and artificial intelligence. Nature Medicine, 25(1), 44-56. https://doi.org/10.1038/s41591-018-0300-7

2. APMG International. (2020). Artificial Intelligence (AI) Foundation Syllabus. Section 1: AI Applications and Benefits.

3. Esteva, A., et al. (2019). A guide to deep learning in healthcare. Nature Medicine, 25(1), 24-29. https://doi.org/10.1038/s41591-018-0316-z

Based on the European Commission's Ethics Guidelines for Trustworthy AI, achieving human-centric, trustworthy AI is not a static, one-time achievement but a continuous, iterative process. The guidelines explicitly state that AI systems must be continually assessed, monitored, and updated throughout their entire lifecycle. This ongoing evaluation ensures that the AI system remains robust, safe, and aligned with ethical principles as it learns, evolves, and interacts with changing environments. Continuous monitoring is essential to identify and mitigate emerging risks, biases, or unintended consequences that may not have been present or visible during the initial design and deployment phases, thereby maintaining public trust and safety.

A. quality assurance certified: While QA is part of software development, the specific ethical framework for Trustworthy AI mandates continuous lifecycle assessment rather than a singular certification.

C. financially sustainable: Financial sustainability is a business objective, not a requirement for an AI system to be classified as ethically trustworthy or human-centric.

D. tested by humans: Human oversight is required, but testing alone is insufficient; the framework specifically demands continuous assessment and monitoring throughout the operational lifecycle.

1. High-Level Expert Group on Artificial Intelligence (AI HLEG). (2019). Ethics Guidelines for Trustworthy AI. European Commission. Chapter 3 (The Assessment List for Trustworthy AI), pp. 26-27.

In computer programming and data science, a vector is fundamentally represented as a one-dimensional array of elements, typically numerical values like integers or floating-point numbers. While in physics or vector calculus, a vector denotes magnitude and direction, programming languages (such as Python, C++, or Java) implement vectors as contiguous memory locations or dynamic arrays accessed via a single index. In the context of Artificial Intelligence and Machine Learning, these one-dimensional arrays are crucial for representing feature sets, weights, or biases in neural networks. They form the foundational data structure upon which higher-order tensors (like matrices) are built, enabling efficient linear algebra computations required for training algorithms.

B. A two-dimensional array of scalars represents a matrix, not a vector, as it requires two indices (row and column) for element access.

C. While a vector can contain complex numbers, defining it strictly as an array of complex numbers is too narrow; vectors typically contain real numbers.

D. A constant represents a single, unchangeable value (a scalar), lacking the multi-element, one-dimensional structure characteristic of a vector.

1. Goodfellow, I., Bengio, Y., & Courville, A. (2016). Deep Learning. MIT Press. Section 2.1: Scalars, Vectors, Matrices and Tensors, p. 31.

2. VanderPlas, J. (2016). Python Data Science Handbook. O'Reilly Media. Chapter 2: Introduction to NumPy, p. 34.

Agile project management methodologies are fundamentally characterized by collaboration, continuous learning, and iterative development cycles. Unlike traditional sequential methods, Agile emphasizes cross-functional teamwork, frequent stakeholder collaboration, and the ability to adapt to changing requirements through empirical learning. The iterative nature allows teams to deliver functional increments of a product, gather feedback, and refine the solution in subsequent cycles. In the context of Artificial Intelligence and software development, Agile frameworks like Scrum facilitate rapid adaptation to complex, evolving data models and algorithms. This approach ensures that learning from each iteration directly informs the next, maximizing value delivery and mitigating risks associated with long-term, rigid project plans.

A. Waterfall projects: Waterfall is a linear, sequential approach, lacking the iterative and continuous learning loops characteristic of Agile methodologies.

B. Rapid software development: While related to speed, RAD focuses more on prototyping than the specific cultural pillars of collaboration, learning, and iterative delivery defining Agile.

C. Trustworthy AI: This refers to AI ethics, fairness, transparency, and reliability, not project management methodologies or development lifecycles.

1. APMG International. (2020). Artificial Intelligence (AI) Foundation Syllabus. Section 4: AI and Agile.

2. Beck, K., et al. (2001). Manifesto for Agile Software Development. Agile Alliance. Available at: https://agilemanifesto.org/

3. Project Management Institute (PMI). (2017). Agile Practice Guide. Section 2.1 (Definable Work vs. High-Uncertainty Work), pp. 7-9.

TensorFlow is an end-to-end open-source machine learning framework and library originally developed by researchers and engineers working on the Google Brain team. It provides a comprehensive, flexible ecosystem of tools, libraries, and community resources that enables researchers to advance the state-of-the-art in machine learning and allows developers to easily build and deploy ML-powered applications. At its core, TensorFlow is a mathematical library based on dataflow and differentiable programming, widely used for training and inference of deep neural networks. It is not restricted to cloud environments, nor is it an intelligent robot paradigm or an agent-based modeling application, making it definitively a machine learning library.

A. TensorFlow is a framework and library that can be deployed anywhere (edge devices, local servers, cloud), not exclusively a cloud-based AI application.

B. It is a software library for numerical computation and machine learning, not a paradigm or framework specifically designed for intelligent robotics.

D. While it can be used to build various models, it is a general-purpose machine learning library, not a dedicated agent-based modeling application like NetLogo.

1. Abadi, M., et al. (2016). "TensorFlow: A System for Large-Scale Machine Learning." 12th USENIX Symposium on Operating Systems Design and Implementation (OSDI 16), pp. 265-283.

2. Official TensorFlow Documentation. "About TensorFlow." TensorFlow.org, Introduction Section.

In an artificial intelligence project, a domain expert (often referred to as a Subject Matter Expert or SME) is an individual possessing deep, specialized knowledge, skills, and practical experience in the specific field or industry for which the AI solution is being developed. Their primary role is to provide essential business context, guide the data selection and feature engineering processes, and validate the machine learning model's outputs against real-world scenarios. Because data scientists may lack specific industry knowledge, the domain expert is crucial for interpreting results and ultimately defining what constitutes a "fit for purpose" solution that accurately solves the problem and meets the actual business or operational requirements of the organization.

A. Technical and managerial oversight of the business plan is typically the responsibility of a project sponsor, product owner, or business manager, not the domain expert.

B. Managing the agile project and writing technical terms of reference are the specific duties of a project manager, scrum master, or technical lead.

C. Measuring the trustworthiness of the AI system involves AI ethicists, quality assurance testers, and compliance officers, though domain experts may assist in evaluating specific outcomes.

- Provost, F., & Fawcett, T. (2013). Data Science for Business. O'Reilly Media. Chapter 2 (Business Problems and Data Science Solutions), pp. 27-30.

- APMG International, Artificial Intelligence (AI) Foundation Syllabus, Section: Building an AI System.

Machine learning enables robot autonomy by allowing the system to process complex, unstructured sensor data (such as vision, lidar, or tactile feedback) and use that data to learn, adapt, and plan actions without human intervention. True autonomy requires an agent to operate successfully in dynamic and unpredictable environments. By applying machine learning algorithms, a robot can build internal models of its environment, recognize patterns, and optimize its decision-making processes over time. This continuous loop of learning from sensory input and dynamically planning tasks allows the robot to achieve its goals independently, moving beyond rigid, pre-programmed instructions to flexible, intelligent behavior.

A. Use OCR to read documents: OCR is a specific application of pattern recognition, but reading documents alone does not grant a robot physical autonomy or task planning capabilities.

B. Use NLP to listen: NLP allows a system to understand human language, which is an interface feature, not the core mechanism of autonomous task planning.

C. Use actuators to modify its environment: Actuators are the physical hardware used to move; they do not involve machine learning or provide the intelligence required for autonomy.

1. Siegwart, R., Nourbakhsh, I. R., & Scaramuzza, D. (2011). Introduction to Autonomous Mobile Robots (2nd ed.). MIT Press. Chapter 1 (Introduction to Autonomy).

2. APMG International. Artificial Intelligence Foundation (AIF) Syllabus. Section: Robotics and Machine Learning.

Universal design is the concept of designing products, environments, programs, and services to be usable by all people, to the greatest extent possible, without the need for adaptation or specialized design. The United Nations Convention on the Rights of Persons with Disabilities (CRPD) explicitly defines and mandates universal design to ensure accessibility and inclusion. In the realm of Artificial Intelligence, applying universal design principles ensures that AI systems and interfaces are accessible to diverse user groups, including those with disabilities, thereby preventing digital exclusion. The European Union also heavily integrates this principle into its accessibility directives and ethical guidelines for trustworthy AI, emphasizing human-centric and inclusive technology.

A. Core design is a generic term and not the recognized, standardized framework for inclusive accessibility mandated by the UN and EU.

C. Biophilic design refers to integrating natural elements into architecture and the built environment to connect humans with nature, not accessibility.

D. Utopic design is a philosophical concept regarding ideal societies, not a practical, legislated design standard for accessibility and inclusion.

1. United Nations. (2006). "Convention on the Rights of Persons with Disabilities (CRPD)", Article 2 (Definitions).

2. European Commission, High-Level Expert Group on AI. (2019). "Ethics Guidelines for Trustworthy AI", Section on Diversity, Non-discrimination and Fairness, p. 18.

3. APMG International. "Artificial Intelligence (AI) Foundation Syllabus", Section on Ethical AI and Human-Centric Design.