Standard

ISO/IEC/IEEE 12207:2017 also provides processes that can be employed for defining, controlling, and improving software life cycle processes within an organization or a project.The processes, activities, and tasks of this document can also be applied during the acquisition of a system that contains software, either alone or in conjunction with ISO/IEC/IEEE 15288:2015, Systems and software engineering?System life cycle processes.In the context of this document and ISO/IEC/IEEE 15288, there is a continuum of human-made systems from those that use little or no software to those in which software is the primary interest. It is rare to encounter a complex system without software, and all software systems require physical system components (hardware) to operate, either as part of the software system-of-interest or as an enabling system or infrastructure. Thus, the choice of whether to apply this document for the software life cycle processes, or ISO/IEC/IEEE 15288:2015, Systems and software engineering?System life cycle processes, depends on the system-of-interest. Processes in both documents have the same process purpose and process outcomes, but differ in activities and tasks to perform software engineering or systems engineering, respectively.

This document specifies the syntax and semantics of description schemes to represent data exchanged by media things (e.g., media sensors, media actuators, media analysers, media storages). Moreover, it specifies the APIs to exchange these data between media things. This document does not specify how sensing and analysing is carried out but defines the interfaces between the media things.

A set of operator interfaces frequently used in artificial intelligence (AI) applications is defined in this standard, where the AI operators refer to the standard building blocks and primitives for performing basic AI operations. The functionality and the specific input and output operands of an AI operator are discussed, as well as both generality and efficiency. Various types of operators, such as those related to basic mathematics, neural network, and machine learning, are highlighted.

This standard defines attributes related to industrial Artificial Intelligence (AI) data that facilitates the classification, association, and mapping towards value creation using data. The attributes include but are not limited to data source, type, ownership, sampling frequency, traceability, privacy attributes for modeling, sampling, shareability and its use in AI algorithms.

This standard defines mandatory and optional requirements and constraints that need to be satisfied for an AI method, algorithm, application or system to be recognized as explainable. Both partially explainable and fully or strongly explainable methods, algorithms and systems are defined. XML Schema are also defined.

This standard establishes a uniform set of definitions and a process model for the procurement of Artificial Intelligence (AI) and Automated Decision Systems (ADS) by which government entities can address socio-technical and responsible innovation considerations to serve the public interest. The process requirements include a framing of procurement from an IEEE Ethically Aligned Design (EAD) foundation and a participatory approach that redefines traditional stages of procurement as: problem definition, planning, solicitation, critical evaluation of technology solutions (e.g. Impact assessments), and contract execution. The scope of the standard not only addresses the procurement of AI in general, but also government in-house development and hybrid public-private development of AI and ADS as an extension of internal government procurement practices.

The standard defines specific terminology utilized in artificial intelligence and machine learning (AI/ML). The standard provides clear definition for relevant terms in AI/ML. Furthermore, the standard defines requirements for data formats.

This recommended practice establishes an evaluation framework for the capabilities of artificial intelligence dialogue systems such as chatbots, consulting terminals, or operation interfaces. The recommended practice defines and classifies the types and levels of the intelligence capabilities according to a checklist of criteria. The checklist tables describe the criteria used to determine the level that a dialogue system achieves based on the analysis of behavior and performance.

Test specifications with a set of indicators for common corruption and adversarial attacks, which can be used to evaluate the robustness of artificial intelligence-based image recognition services are provided in this standard. Robustness attack threats and establishes an assessment framework to evaluate the robustness of artificial intelligence-based image recognition service under various settings are also specified in this standard.

This standard describes recognizable audio and visual marks to assist with the identification of communicating entities as human or machine intelligence to facilitate transparency, understanding, and trust during online, telephone, or other electronic interactions. Interventions to discern whether an interaction is with a machine or not (such as a Turing Test) are not within the scope of this standard. This standard is concerned only about the declaration of the nature of the agency influencing an interaction.

To coordinate global data and AI literacy building efforts, this standard establishes an operational framework and associated capabilities for designing policy interventions, tracking their progress, and empirically evaluating their outcomes. The standard includes a common set of definitions, language, and understanding of data and AI literacy, skills, and readiness.

This standard describes specific methodologies to help users certify how they worked to address and eliminate issues of negative bias in the creation of their algorithms, where negative bias infers the usage of overly subjective or uniformed data sets or information known to be inconsistent with legislation concerning certain protected characteristics (such as race, gender, sexuality, etc); or with instances of bias against groups not necessarily protected explicitly by legislation, but otherwise diminishing stakeholder or user well being and for which there are good reasons to be considered inappropriate. Possible elements include (but are not limited to): benchmarking procedures and criteria for the selection of validation data sets for bias quality control; guidelines on establishing and communicating the application boundaries for which the algorithm has been designed and validated to guard against unintended consequences arising from out-of-bound application of algorithms; suggestions for user expectation management to mitigate bias due to incorrect interpretation of systems outputs by users (e.g. correlation vs. causation).