AI Assessment as a Building Block for Operationalizing the AI Act – Experience Reports from Pilot Projects
Day 2 | 11:30 – 12:00 | Workshop Room 2
Anna Schmitz
Fraunhofer IAIS
Abstract
With the adoption of the European AI Regulation, relevant regulatory requirements for the use of AI come into force. It is now important for companies to use the transition period to implement the regulatory requirements efficiently and pragmatically and to demonstrate conformity with them. Also in unregulated domains, quality proofs are an important factor in building trust in AI systems. This presentation gives a brief introduction to current approaches to AI assessment and outlines relevant international frameworks and standards. In particular, it distinguishes between different assessment objects i.e., the organization and the AI system. As many requirements are still abstract, it is important to learn from practical experience. Therefore, the presentation provides an insight into the experiences from pilot assessments and lessons learned that may be valuable for other companies.
Anna Schmitz
Anna Schmitz works at the Fraunhofer-Institute for Intelligent Analysis and Information Systems as a scientist and doctoral candidate in the field of trustworthy AI and assessment of AI applications. She is particularly concerned with the question of how regulatory requirements for AI systems can be implemented in practice and how conformity can be demonstrated. In current projects, she is investigating the standardization landscape for AI and developing systematic approaches to risk analysis and assessment, which are piloted on real AI systems. In addition to advising companies and public authorities on these topics, she also gives seminars on trustworthy AI.