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
Here is an overview of published international standards of the DIN/DKE Joint Working Committee on Artificial Intelligence.
Here is an overview of current national standards of the DIN/DKE Joint Working Committee on Artificial.
Ongoing standardization projects
Artificial intelligence - Quality requirements and processes - Risk analysis for the development and operation of AI systems
DIN/TS 92004 contains requirements for risk analysis and treatment for the development and operation of artificial intelligence (AI) systems that contain machine learning (ML) components. It specifies the assessment of AI-relevant safety-related risks for these systems and considers application-specific risks with regard to fairness, autonomy, transparency and privacy.
Prevention of food waste - Digital reporting of food surpluses in the supply chain - Part 1: Definition of semantics and a data protocol for digital transmission
DIN SPEC 91550-1 pursues the goal of avoiding food waste in the supply chain through rapid and data-driven distribution of product surpluses to relevant acceptance channels. For a technology-driven elimination of food waste in the supply chain, the surpluses of the supply chain partners can be allocated by means of an increasingly AI-driven distribution platform. This requires a standardized reporting protocol with equally standardized semantics to exchange surplus data between the platform and the mentioned actors.
Artificial intelligence - Life cycle processes and quality requirements - Part 3: Explainability
DIN SPEC 92001-3 is intended to provide an industry-independent guide to appropriate approaches and methods for promoting explainability throughout the life cycle of an AI model. The quality criteria described in this document aim to be applicable to all types of AI systems and to all domains in which such systems are used. In particular, they apply to AI systems with varying degrees of opacity.
Artificial intelligence - Quantification of uncertainties in machine learning
DIN SPEC 92005 specifies general guidance and requirements for the development and use of methods for quantifying uncertainty in machine learning (ML). defines This standard defines basic terms for quantifying uncertainty for ML and specifies the purpose, use, and need for these analyses. Likewise, this document provides an overview of existing technical methods of uncertainty quantification for ML and their properties, and describes selected applications.