ISTQB - AI Testing(CT-AI) - 1. Introduction to AI -Test 01

Which of the following best describes the "AI effect"?. The point at which AI systems surpass human intelligence. The capability of an engineered system to acquire, process, and apply knowledge and skills. The changing perception of what constitutes AI as technology evolves and previously impressive feats become commonplace. The method by which AI systems learn from vast amounts of data.

How is "Artificial Intelligence (AI)" defined in the syllabus?. The programming of machines to perform tasks that typically require human intelligence, such as visual perception and decision-making. The capability of an engineered system to acquire, process and apply knowledge and skills. A field of study focused on creating machines that can perfectly replicate human emotional responses. The use of advanced algorithms to solve complex mathematical problems.

What primarily influences the evolving definition of AI?. The processing power of available hardware. The changing perception in society of what Al is. The amount of funding allocated to AI research. The number of programming languages available for AI development.

What is a key characteristic of "Narrow AI" (weak AI)?. It possesses general cognitive abilities similar to humans. It is programmed to carry out a specific task with limited context. It is a theoretical concept that has not yet been realized. It can replicate human cognition and utilize unlimited memory.

Which statement accurately describes "General AI" (strong AI)?. It is widely available in everyday applications like spam filters. It has general (wide-ranging) cognitive abilities similar to humans. It is primarily used for playing complex games like chess. As of 2021, general AI systems are commonly used in industrial automation.

What is "Super AI"?. AI systems that are highly specialized in one narrow task. AI systems that can replicate human cognition and make use of massive processing power and unlimited memory, expected to become wiser than humans. The current state-of-the-art in AI, available as a cloud service. AI that operates exclusively on mobile devices.

What is the "technological singularity"?. The point at which AI-based systems can no longer be understood by humans. The moment a single AI system can perform all human tasks. The point at which AI-based systems transition from general AI to super AI. The development of the first AI programming language.

How does a typical conventional computer system primarily operate?. By learning patterns from large datasets. The software is programmed by humans using an imperative language with constructs like if-then-else. It determines on its own what patterns or features in data can be used. Its prediction-making procedure is often difficult for humans to understand.

What is a defining characteristic of an AI-based system using machine learning (ML)?. It is always easier for humans to understand its input-output transformation compared to conventional systems. It relies solely on explicitly programmed rules by humans. Patterns in data are used by the system to determine how it should react in the future to new data. It cannot be implemented using technologies like neural networks.

Which of the following is listed as a technology used to implement AI?. SQL databases. HTTP protocols. Fuzzy logic. Integrated Development Environments (IDEs).

Which of these is NOT a machine learning technique mentioned in the syllabus?. Decision trees. Blockchain encryption. Support vector machine (SVM). Clustering algorithms.

What is TensorFlow?. A type of neuromorphic processor. An open-source ML framework based on data flow graphs for scalable machine learning, provided by Google. A standard for ensuring ethical AI development. A low-precision arithmetic unit.

Which statement is true about Keras?. It is a deep learning open-source framework used by Amazon for AWS. It is an open-source ML library operated by Facebook. It is a high-level open-source API, written in Python, capable of running on top of TensorFlow and CNTK. It is a suite of tools that support the development of AI solutions provided by IBM.

Which hardware attribute is generally less critical for ML applications compared to others listed?. Low-precision arithmetic. Ability to work with large data structures. Support for complex general-purpose operations with few cores. Massively parallel (concurrent) processing.

Why do GPUs typically outperform CPUs for ML applications?. CPUs have thousands of cores designed for massively parallel processing. GPUs are designed for massively parallel processing and have thousands of cores. CPUs generally have lower clock speeds than GPUs. GPUs use higher precision arithmetic which is better for ML.

What are neuromorphic processors?. Standard CPUs with enhanced clock speeds. Processors that do not use the traditional von Neumann architecture, but rather an architecture that loosely mimics brain neurons. A type of GPU specialized for image rendering. Cloud-based tensor processing units.

What is "AI as a Service (AIaaS)"?. A hardware specification for building AI-compatible computers. A method where AI components, like ML models, are used as a service over the web. A programming language specifically designed for AI development. The practice of training AI models exclusively on local machines.

What is a typical benefit of using AIaaS for organizations. It guarantees full liability coverage from the service provider for any AI malfunction. It allows implementation of AI using cloud-based services even with insufficient internal resources or expertise. It always results in greater transparency compared to in-house developed models. It eliminates the need for testing the AI service.

What is a common characteristic of Service Level Agreements (SLAs) for AIaaS. They primarily define ML functional performance metrics like accuracy. They typically cover uptime for the service and response time to fix defects. They usually offer unlimited liability for the service provider. They mandate the use of specific hardware by the consumer.

What is a key advantage of using pre-trained models?. They always perfectly match the new required functionality without any modification. Training costs are saved and the risk of the model not working is largely eliminated. They are guaranteed to be free of any biases. They offer greater transparency than internally generated models.

What is "transfer learning"?. The process of transferring a trained model from one cloud provider to another. Taking a pre-trained model and modifying it to perform a second, different requirement, often reusing early layers. A technique for learning exclusively from small datasets. A method to convert conventional software into an AI-based system.

Which of the following is a risk associated with using pre-trained models?. They are always more expensive than developing models from scratch. Inherited biases may not be apparent if there is a lack of documentation about the training data. They require complete retraining for any new task. They are less likely to be vulnerable to adversarial attacks.

What is a potential risk of using transfer learning?. It makes models completely immune to vulnerabilities of the original pre-trained model. It always requires significantly more data than training a model from scratch. Models created through transfer learning are highly likely to be sensitive to the same vulnerabilities as the pre-trained model. It guarantees full documentation of the original model's shortcomings.

Which joint technical committee of IEC and ISO prepares international standards contributing to AI?. ISO/IEC JTC 1/SC7. ISO/IEC JTC 1/SC42. IEEE Global Initiative. The German national standards body (DIN).

What is a key aspect of the GDPR from a testing perspective for AI systems?. All AI predictions must be 100% accurate. Personal data (including predictions) should be accurate enough for the purposes for which it is used. AI systems must exclusively use open-source pre-trained models. Data controllers are exempt from obligations regarding automated decision-making.

Which form of AI is described as being "widely available" as of 2021?. General AI. Super AI. Narrow AI. Theoretical AI.

Which of the following is an example of an AI development framework?. GDPR. PyTorch. Technological Singularity. SVM (Support Vector Machine).

According to the syllabus, which of the following is NOT an example of AlaaS?. IBM Watson Assistant. Google Cloud AI and ML Products for document AI. A locally installed Python compiler. Amazon CodeGuru.

What does the syllabus state about the relationship between the AI Effect and the definition of an AI-based system?. The AI Effect stabilizes the definition of an AI-based system over time. The AI Effect may determine what is currently considered to be an AI-based system versus a conventional system. The AI Effect is irrelevant to how AI-based systems are defined. The AI Effect only applies to Super AI systems.

What hardware feature involves using fewer bits for computation (e.g., 8 instead of 32 bits), which is usually sufficient for ML?. Massively parallel processing. In-memory processing. Low-precision arithmetic. High clock speed.