German Standardization Roadmap on Artificial Intelligence
2 edition, (2022)Abstract
On behalf of the German Federal Ministry of Economic Affairs and Climate Action, DIN and DKE started work on the second edition of the German Standardization Roadmap Artificial Intelligence in January 2022. With the broad participation and involvement of more than 570 experts from industry, science, the public sector and civil society, the strategic Roadmap for AI standardization was thus further developed. This work was coordinated and accompanied by a high-level coordination group for AI standardization and conformity
The standardization roadmap implements a measure of the German government’s AI Strategy and thus makes a significant contribution to “AI – Made in Germany”.
Standardization is part of the AI Strategy and is a strategic instrument for strengthening the innovation and competitiveness of the German and European economies. Not least for this reason, standardization plays a special role in the planned European legal framework for AI, the Artificial Intelligence Act.
This Standardization Roadmap AI identifies the requirements in standardization, formulates concrete recommendations and thus creates the basis for initiating standardization work at national level, and especially at European and international level, at an early stage. In doing so, the Roadmap makes a significant contribution to the European Commission’s Artificial Intelligence Act, supporting its implementation.
The Standardization Roadmap AI focuses on nine key topics, which are addressed in Chapter 4:
→ The Roadmap begins with the basic topics, such as terminologies and definitions, classifications and ethical issues. They are the basis for AI discussions and are thus the central core of the Roadmap.
→ The security/safety of AI systems plays a crucial role in widespread use of AI solutions. Only a more in-depth consideration of requirements for operational safety and information security, for example, can enable the comprehensive use of AI systems in business and society.
→ Another key topic, and the basis for the broad market success of AI, is testing and certification. This requires reliable quality criteria and reproducible test methods that can be used to verify the properties of AI systems. They are a key prerequisite for assessing the quality of AI-based applications and contribute significantly to explainability and traceability – two factors that build trust and acceptance.
→ Another challenge in the use of AI, especially for small and medium-sized enterprises, is the integration of AI technologies in organizations. The focus here is on sociotechnical aspects such as human-technology interaction, humane work design, and requirements for business structures and processes, which are all examined in the Roadmap.
→ The fields of application of AI are extremely diverse. AI technologies are used in almost all business and application areas and offer great potential. To cover a broad spectrum of applications, the Roadmap considers industry-specific challenges for the following five sectors in particular, in addition to the cross-cutting issues mentioned above: Industrial Automation, Mobility, Medicine, Financial Services and Energy / Environment.
The present Roadmap outlines the work and discussion results for all nine key topics and provides a comprehensive overview of the status quo, requirements, and needs for action.