Standards and specifications might help to ensure that artificial intelligence works safely and dependably. This concerns in particular the quality, security, traceability, robustness, transparency and reliability of AI applications. Standardization enables German SMEs in particular to gain access to the global market via open interfaces and uniform requirements.
On national as well as on international level the first standards and specifications for artificial intelligence have already been developed.
Current standards and specifications on artificial intelligence
Guideline for the development of deep learning image recognition systems
DIN SPEC 13266 provides basic knowledge about the application possibilities and the structure of deep learning systems, and names the conditions under which image recognition problems can be considered by using a deep learning system. Furthermore, the document provides guidelines for practical implementation - including the collection of data, their structuring, the structure of learning experiments and error analysis.
Guideline for the development of deep learning image recognition systems in medicine
Similar to DIN SPEC 13266:2020, DIN SPEC 13288 is intended to provide a guide to the increased AI-specific quality requirements and regulatory specifications in the application of medical image recognition systems. Starting with a representative data set through the embedding in the medical workflow and the applied methodology, explainability and user confidence. DIN SPEC 13288 also contains aspects of technical implementation: for example, for formalization/problem definition, software development or continual improvement and validation.
Artificial Intelligence — Life Cycle Processes and Quality Requirements — Part 1: Quality Meta Model
Part 1 of the DIN SPEC 92001 series provides a general quality meta model for artificial intelligence (AI), which primarily describes the most important aspects of AI quality. The AI quality meta model contains, among other things, the three essential quality characteristics - performance & functionality, robustness and comprehensibility.
Artificial Intelligence — Life Cycle Processes and Quality Requirements — Part 2: AI robustness
Part 2 of the DIN SPEC 92001 series specifies AI-specific robustness requirements. These quality requirements are structured using the specified AI quality meta model (DIN SPEC 92001-1).
Quality requirements for video-based methods of personnel selection
DIN SPEC 91426 contains a practical personnel diagnostic and legal framework for video-based methods of personnel selection.
Transmission of language-based data between artificial intelligences - Specification of parameters and formats
DIN SPEC 2343 is intended to define a universal grammar for language interfaces, which enables different AI ecosystems to be brought together within a common language framework and ensures interoperability.
Information technology - Big data - Overview and vocabulary
ISO/IEC 20546 provides a set of terms and definitions needed to promote improved communication and understanding of this area. It provides a terminological foundation for big data-related standards.
This document provides a conceptual overview of the field of big data, its relationship to other technical areas and standards efforts, and the concepts ascribed to big data that are not new to big data.
Information technology - Big data reference architecture - Part 3: Reference architecture
ISO/IEC 20547-3 - The BDRA is intended to facilitate aspects such as common language ,encourage adherence to common standards, specifications and patterns, and the analysis of candidate standards for interoperability, portability, reusability, and extendibility.
Information technology - Big data reference architecture - Part 1: Framework and application process
ISO/IEC TR 20547-1 describes the framework of the big data reference architecture and the process for how a user of the document can apply it to their particular problem domain.
Information technology - Big data reference architecture - Part 2: Use cases and derived requirements
ISO/IEC TR 20547-2 provides examples of big data use cases with application domains and technical considerations derived from the contributed use cases.
Information technology - Big data reference architecture - Part 5: Standards roadmap
ISO/IEC TR 20547-5 describes big data relevant standards, both in existence and under development, along with priorities for future big data standards development based on gap analysis.
Information technology - Artificial intelligence - Overview of trustworthiness in artificial intelligence
ISO/IEC TR 24028 surveys topics related to trustworthiness in AI systems, including for example approaches to establish trust in AI systems through transparency, explainability and controllability, engineering pitfalls and typical associated threats and risks to AI systems, along with possible mitigation techniques and methods; and approaches to assess and achieve availability, resiliency, reliability, accuracy, safety, security and privacy of AI systems.