Visual inspection is a core AI application used in quality assurance processes. However, in some sensitive use cases, such as pharma, transparency about decision is demanded and humans are responsible for taking decisions finally. Hence, humans must rely on AI and also trust AI recommendations. XAI can provide a certain level of transparency and is a promising avenue to foster trust. The question however is:
Which type of XAI should be presented in which form to enable a sufficient human – AI collaboration?
Task description
- Literature research XAI in regulated areas
- Conduct a case study in quality assurance of Fresenius Kabi (Graz) to elicit requirements and context factors for successful XAI decision support
Supervision
- Scientific Supervisor : Univ.-Prof. Dr. Stefan Thalmann
- Co- Supervisor : Dr. Jürgen Fleiß + Dipl.-Ing. Christoph Koeth - Fresenius Kabi