The term is discussed in various disciplines from different perspectives.  Due to the AI effect, the term is constantly evolving. Three definitions are given below:

Definition 1: Capability of an engineered system to acquire, process and apply knowledge and skills (ISO/IEC 29119-11).

Definition 2: Computerized system that uses cognition to understand information and solve problems (ISO/IEC 22989).

Definition 3: Artificial intelligence refers to a set of technologies [...] that, for a given set of human-defined objectives, generate outputs such as content, predictions, recommendations, or decisions influencing the environments they interact with (European AI Act in draft ¹).

• Machine Learning
• AI effect

ETSI GR ENI 004 computerized system that uses cognition to understand information and solve problems

NOTE 1: ISO/IEC 2382-28 defines AI as "an interdisciplinary field, usually regarded as a branch of computer science, dealing with models and systems for the performance of functions generally associated with human intelligence, such as reasoning and learning".

NOTE 2: In computer science AI research is defined as the study of "intelligent agents": any device that perceives its environment and takes actions to achieve its goals.

NOTE 3: This includes pattern recognition and the application of machine learning and related techniques.

NOTE 4: Artificial Intelligence is the whole idea and concepts of machines being able to carry out tasks in a way that mimics the human intelligence and would be considered "smart".

ISO/IEC DIS 22989 set of methods or automated entities that together build, optimize and apply a model (3.1.26) so that the system can, for a given set of predefined tasks (3.1.37), compute predictions (3.2.12), recommendations, or decisions.

Note 1 to entry: AI systems are designed to operate with varying levels of automation (3.1.7).

Note 2 to entry: Predictions (3.2.12) can refer to various kinds of data analysis or production (including translating text, creating synthetic images or diagnosing a previous power failure). It does not imply anteriority. study of theories, mechanisms, developments and applications related to artificial intelligence (3.1.2).

ISO/IEC TR 29119-11:2020 capability of an engineered system to acquire, process and apply knowledge and skills

ISTQB - CTAI Syllabus The capability of an engineered system to acquire, process, create and apply knowledge and skills (ISO/IEC TR 29119-11)

Software module in which AI methods are implemented.

The case when an AI system is no longer considered to be AI due to the advancement of technology.

ISO/IEC TR 29119-11:2020 situation when a previously labelled AI (3.1.13) system is no longer considered to be AI as technology advances

ISTQB - CTAI Syllabus The situation when a previously labelled AI system is no longer considered to be AI as technology advances (ISO/IEC TR 29119-11)

A system that integrates one or more AI components.

ISO/IEC TR 29119-11:2020 system including one or more components implementing AI (3.1.13)

ISTQB - CTAI Syllabus A system that integrates one or more AI components

A component that involves AI methods.

• component

ISTQB - CTAI Syllabus A component that provides AI functionality

A system that uses Artificial Intelligence and consists of several components (at least one of which is an AI component).

The AIA¹ compromise proposal adopted by the EU Parliament on 27 April 2023 for the trilogue (based on document version of 06 December 2022) defines an AI system as a system (in contrast to the OECD definition²: there as a ''machine-based system'') that makes predictions, recommendations, or can make decisions that influence the real or virtual environment for certain human-defined goals. AI systems can be endowed with varying degrees of autonomy.

• AI component

A software model for the provision of AI services.

ISTQB - CTAI Syllabus A software licensing and delivery model in which AI and AI development services are centrally hosted

Capability such as percept, process, act and communicate, which can be realised on the basis of AI methods (see DIN/DKE Standardisation Roadmap on AI, 2. Edition, section 4.1.2.1).

Metamodel for assuring the quality of AI-based systems.
Note: This metamodel is defined in detail in DIN SPEC 92001.

ISO/IEC TR 29119-11:2020 metamodel intended to ensure the quality of AI-based systems (3.1.9)

Note 1 to entry: This metamodel is defined in detail in DIN SPEC 92001.

Data used to evaluate the performance of a final AI model (in general) or machine learning model (in particular) before it goes live.

Note: Basically, test data shall be disjoint from training data and validation data.

ISO/IEC DIS 22989 data used to assess the performance of a final machine learning model (3.2.11).

Note 1 to entry: Test data is disjoint from training data (3.2.22) and validation data (3.2.24).

ISO/IEC TR 29119-11:2020 independent dataset used to provide an unbiased evaluation of the final, tuned ML model (3.1.46)

AI that exhibits intelligent behaviour comparable to a human across the full range of cognitive abilities (synonym: strong AI).

• narrow AI

ISO/IEC DIS 22989 AI that addresses a broad range of tasks (3.1.37) with a satisfactory level of performance Note 1 to entry: Compared to narrow AI (3.1.27).

ISO/IEC TR 29119-11:2020 strong AI: AI (3.1.13) that exhibits intelligent behaviour comparable to a human across the full range of cognitive abilities

ISTQB - CTAI Syllabus AI that exhibits intelligent behaviour comparable to a human across the full range of cognitive abilities (ISO/IEC TR 29119-11) Synonym: strong AI

AI focused on a single well-defined task to address a specific problem.

• general AI

ISO/IEC DIS 22989 AI (3.1.13) focused on a single well-defined task to address a specific problem.

ISO/IEC TR 29119-11:2020 weak AI: AI (3.1.13) focused on a single well-defined task to address a specific problem

ISTQB - CTAI Syllabus AI focused on a single well-defined task to address a specific problem (ISO/IEC TR 29119-11)

Synonym: weak AI

A research and application area concerned with understanding the factors that influence the outcomes of AI systems.

ISTQB - CTAI Syllabus The field of study related to understanding the factors that influence AI system outputs

A Type of AI method based on models that uses symbols and structures to recognize patterns.

• subsymbolic AI

ISO/IEC DIS 22989 AI (3.1.2) based on techniques and models (3.1.26) using symbols and structures

Note 1 to entry: Compared to. subsymbolic AI (3.1.36).

Type of AI Methods based on models using a numeric representation and implicit information encoding.

• symbolic AI

ISO/IEC DIS 22989 AI (3.1.2) based on techniques and models (3.1.26) using a numeric representation and implicit information encoding

Note 1: Compared to symbolic AI (3.1.35).

Computational linguistics studies how natural language in the form of text or language data can be processed algorithmically with the help of computers. It represents the interface between linguistics and computer science.

An application-specific integrated circuit designed to manipulate and alter memory to accelerate the creation of images in a frame buffer intended for output to a display device.

ISO/IEC TR 29119-11:2020 GPU: application-specific integrated circuit (ASIC) specialized for display functions such as rendering images

Note 1 to entry: GPUs are designed for parallel data processing of images with a single function, but this parallel processing is also useful for executing AI-based software, such as neural networks (3.1.48).

ISTQB - CTAI Syllabus An application-specific integrated circuit designed to manipulate and alter memory to accelerate the creation of images in a frame buffer intended for output to a display device

Genetic algorithms refer to a method area of machine learning simulating natural selection. In this process, parameter constellations are iteratively modified (mutation) and tested in different combinations (recombination). The goal is to find the optimal combination of parameters to solve a task.

ISO/IEC DIS 22989 algorithm which simulates natural selection by creating and evolving a population of individuals (solutions) for optimization problems

An ML algorithm that groups similar data points together.

ISTQB - CTAI Syllabus A type of ML algorithm used to group similar objects into clusters

Node in an artificial neural network (ANN) that receives weighted input data and subsequently generates outputs based on usually non-linear functions.

ETSI GR ENI 004 node in an ANN that receives weighted input data, adds the data, and produces an output using a non-linear output function.

NOTE: The output function is also called a transfer function or activation function, and represents what portion of the potential action is transmitted.

ISO/IEC DIS 22989 primitive processing element which takes one or more input values and produces an output value by combining the input values and applying an activation function on the result.

Note 1 to entry: Examples of nonlinear activation functions are a threshold function, a sigmoid function and a polynomial function.

ISTQB - CTAI Syllabus neuron: A node in a neural network, usually receiving multiple input values and generating an activation value

A procedure/system used to implement a classification task.

ETSI GR ENI 004 procedure that predicts which elements of a set belong to which (pre-defined) classes

ISO/IEC TR 29119-11:2020 ML model (3.1.46) used for classification (3.1.20)

ISTQB - CTAI Syllabus An ML model used for classification Synonym: classification model

In the context of AI, continual learning is the training of an AI system that takes place iteratively and incrementally simultaneously with its operation.

ISO/IEC DIS 22989 incremental training of an AI system (3.1.4) that takes place on an ongoing basis during the operation phase of the AI system life cycle.

Property of an AI system, describing that a human or other external agent can immediately and without delay intervene in the functioning of the system.

ISO/IEC DIS 22989 controllable property of an AI system (3.1.4) that a human or other external agent (3.1.1) can intervene in the system’s functioning

Note 1 to entry: Such a system is heteronomous (3.1.9).

The information provided by direct observation and measurement that is known to be real or true.

ISO/IEC DIS 22989 value of the target variable for a particular item of labelled input data

Note 1 to entry: Ground truth is not always the same as absolute truth.

ISTQB - CTAI Syllabus The information provided by direct observation and measurement that is known to be real or true

Estimation of an output based on the outputs of the data points closest to it.

ISTQB - CTAI Syllabus An approach to classification that estimates the likelihood of group membership for a data point dependent on the group membership of the data points nearest to it

Closed-box testing (also called "black box testing") means testing of the AI-system in which the principal information used as the basis for designing and implementing test cases is the external inputs and the AI-system's output. Usually, the tester has no access to internals such as the AI model.

ISO/IEC TR 29119-11:2020 testing (3.131) in which the principal test basis (3.84) is the external inputs and outputs of the test item (3.107), commonly based on a specification, rather than its implementation in source code or executable software[ISO/IEC 29119-1:2022, 3.75]

Extent to which a system, product or service can be used by specified users to achieve specified goals with effectiveness, efficiency and satisfaction in a specified context of use.

Note 1: The “specified” users, goals and context of use refer to the particular combination of users, goals and context of use for which usability is being considered.

Note 2: The word “usability” is also used as a qualifier to refer to the design knowledge, competencies, activities and design attributes that contribute to usability, such as usability expertise, usability professional, usability engineering, usability method, usability evaluation, usability heuristic.

cf.: DIN EN ISO 9241-210:2020

In a broad sense, ''weights'' are parameters of a model, usually factors that individually scale (''weight'') specific entries of multidimensional inputs.In artificial neural networks, for example, weights are used to scale input values of an artificial neuron.In machine learning, typically the weights of a model are trained.

ISTQB - CTAI Syllabus An internal variable of a connection between neurons in a neural network that affects how it computes its outputs and that changes as the neural network is trained

A mathematical model that represents connection structures in an abstract fashion. It consists of ''nodes'' and of ''edges'', where the edges represent connections between nodes. In a specific application, nodes as well as edges may be assigned values, such as weights, costs or distances.

ETSI GR ENI 004 collection of nodes, where some subset of the nodes is connected.

NOTE: Visually, a node is a "point" and a connection is a "line", called an "edge". For the purposes of ENI, any graph may be directed, weighted or both.

ANNs are networks of artificial neurons and have a biological blueprint. Leaning on biology, an artificial neuron is an object that reacts to one or more stimuli, depending on how strongly it is activated or the stimulus is weighted. An ANN basically consists of an input layer and an output layer. In between are hidden layers or activity layers. As a rule, ANNs need to be trained before they can solve problems. In the process, a certain algorithm or the neural network weights the connections of the neurons on the basis of predefined learning material and learning rules until it has reached or developed a certain learning goal.

• feed forward neural network (FFNN)
• recurrent neural network (RNN)
• deep neural network (DNN)

ETSI GR ENI 004 computing system that learns to perform functions by using artificial neurons that take the form of a directed, weighted graph.

NOTE: An ANN learns to perform a function by analysing examples (i.e. training data) instead of being programmed to perform a task.

AI capability of a functional unit to acquire, process and interpret visual data. Computer vision involves the use of sensors to create a digital image of a visual scene.

ISO/IEC DIS 22989 capability of a functional unit to acquire, process and interpret visual data

Note 1 to entry: Computer vision involves the use of sensors to create a digital image of a visual scene.

A function-fulfilling element of a system.

• AI component

ETSI GR ENI 004 part of a System that has operational and/or management significance

Confusion matrices are used to evaluate potential weaknesses in classification methods. For a method designed to distinguish N classes, the confusion matrix as a size of  N × N. The numeric value at an element (i, j) denotes how often the method classified an instance of class i as an instance of class j. An ideal method achieves a perfect diagnoal matrix. Off-diagonal entries instead (i.e. cases where for any i≠j the numeric value at this element is nonzero) indicate a risk of the method confusing the classes i and j, and quantify the occurrence.

• true positive
• true negative
• false positive
• false negative

ISO/IEC TR 29119-11:2020 table used to describe the performance of a classifier (3.1.21) on a set of test data (3.1.75) for which the true and false values are known

ISTQB - CTAI Syllabus A technique for summarizing the ML functional performance of a classification algorithm

Understanding data and information and generating new data, information as well as new knowledge.

• cognitive system

ETSI GR ENI 004 process of understanding data and information and producing new data, information and knowledge

Adaptive system with interfaces to the digital world and the environment that can perceive things by itself, can relate them to contexts and understand them, can draw conclusions and learn, all in order to solve and master tasks.

• cognition

Glassbox testing (also called ''white box testing'') means testing of the AI system in which the information used as the basis for designing and implementing test cases can be derived from internal information (e.g. the AI model) as well as the external inputs and the corresponding AI system's output.

ISO/IEC TR 29119-11:2020 dynamic testing (3.29) in which the tests are derived from an examination of the structure of the test item (3.107) (SOURCE: ISO/IEC 29119-1:2022, 3.80)

Measure of potential hazards that may arise from the use of an AI system in a specific application context. The term is often used in a similar way to risk, with criticality focusing more on an assessment of the overall system.

Note: see „Kritikalität, Kritikalitätspyramide, risikoadaptierter Regulierungsansatz“ in „Gutachten der Datenethikkommission“ 10/2019¹ and „Abschlussbericht Enquete-Kommission KI“ 10/2020².

• risk

Concerning classification in the field of AI, accuracy is a metric for measuring the quality of mostly binary classifications. It is calculated as the proportion of correct classifications in all classifications.

• quality measure

ISO/IEC TR 29119-11:2020 performance metric used to evaluate a classifier (3.1.21), which measures the proportion of classifications (3.1.20) predictions (3.1.56) that were correct

ISTQB - CTAI Syllabus The ML functional performance metric used to evaluate a classifier, which measures the proportion of predictions that were correct (After ISO/IEC TR 29119-11)

Task by means of which the output class for a given input is predicted.

If there are only two output classes, this is referred to as binary classification.

If there are more than two output classes, this is referred to as multi-class classification.

ISO/IEC TR 29119-11:2020 machine learning function that predicts the output class for a given input

ISTQB - CTAI Syllabus A type of ML function that predicts the output class for a given input (After ISO/IEC TR 29119-11)

A cognitive architecture is a system that simulates human thought and human thinking and decision-making.

ETSI GR ENI 004 system that learns, reasons, and makes decisions in a manner resembling that of a human mind

NOTE: Specifically, the learning, reasoning, and decision-making is performed using software that makes hypotheses and proves or disproves them using non-imperative mechanisms that typically involve constructing new knowledge dynamically during the decision-making process.