Annual Report
2023 / 2024
2023 / 2024
Leveraging benefits, avoiding risks
There is no question that artificial intelligence (AI) is an absolute game changer. It is now being used in all markets and sectors, and it can save lives – for instance, on the road. But AI is certainly not perfect. The key task is to learn how to harness the benefits of AI while avoiding the risks, particularly in the transportation of the future.
AI is playing an increasingly important role in connected driving and inside the vehicles themselves. What are the biggest challenges this presents from a safety/security perspective?
AI is a major driver of new and personalized customer experiences, but it also introduces uncertainties and new risks. While many AI-based applications such as personalized assistants or driver assistance functions work well in a quick demo setting, making these applications work reliably at scale is a big challenge. The automotive industry must ensure that the AI-based components work as safely as intended, and they have to provide appropriate evidence of that. We believe that our solution helps make the AI do what it’s supposed to do in an automotive setting, while also fulfilling the legal and compliance requirements.
What concepts and solutions is CertifAI looking to employ to overcome these challenges?
We carry out a risk assessment in which we identify the AI component’s risks in the specific use case. Then we define objectives and tests to mitigate these risks. Combined with the development approach based on the 3-V model (Gracic et al., 2020), we can come to a bottom-up conclusion that the residual risk is acceptable and support the overall system safety argumentation. By clearly identifying the risks of the AI system and mitigating them appropriately, automotive customers can develop their AI systems much more efficiently and increase their chances of actually integrating them in a standard operating procedure (SOP).
CertifAI is part of Mission KI, an AI initiative run by Germany’s National Academy of Science and Engineering (Acatech) and Federal Ministry for Digital and Transport (BMDV). Why is this initiative so important, and what contribution does CertifAI make?
CertifAI is part of an expert group defining the testing criteria for a future “AI Made in Germany” seal. The “Made in Germany” mark has been a source of trust in the quality of products around the world, and we’re working hard to establish “AI Made in Germany” with a similar reputation. We believe that making AI do what it’s supposed to do is this century’s most important challenge. Together with our partners, we’ll determine ways to ensure reliable AI development and high-quality AI-based products.
This research and development initiative is one of the flagship projects within the German government’s digital strategy and is intended to strengthen the German economy’s digital competitiveness. Mission KI addresses the challenges of AI by expanding the pool of data for AI innovations and promoting the development and growth of trustworthy, marketable AI applications.
CertifAI was founded by DEKRA, auditing and consulting firm PwC Germany, and Innovationsstarter Fonds Hamburg, a startup fund run by the Hamburg city government. The joint venture’s focus is on providing services in testing and certifying AI models, products, and solutions in compliance with industry-specific norms and standards. The three founding partners want to build trust in AI and meet regulatory demands for technology-driven companies as efficiently as possible. CertifAI’s management is highly experienced in building digital business models in regulated environments.
