Advancing Explainable AI in Manufacturing: Insights from AI-PROFICIENT
The AI-PROFICIENT project is driving the integration of artificial intelligence (AI) into the manufacturing process, while ensuring transparency and explainability. In our recently published deliverable, D4.4: AI-PROFICIENT approach for XAI, authors Dea Pujic, Marc Anderson, Eduardo Gilabert, and Laritza Limia Fernandez shed light on the crucial task of providing Explainable AI (XAI) within the project.
This deliverable serves as a bridge between the technical and ethical aspects of AI-PROFICIENT, with a particular focus on enhancing the acceptance and adoption of AI in production processes. Recognizing the importance of explainability and transparency, the AI-PROFICIENT team dives into the state-of-the-art analysis of various XAI approaches.
In specific, the authors highlight three XAI services developed within the project: the surrogate explainable data-driven model, the post-hoc explainable analysis module, and the auditability system. Detailed methodologies, results, integration, and applications are provided for each service. These XAI services will be implemented in multiple use cases, including CONTI2, CONTI5, CONTI10, and INEOS3.
We must point out that the advancements made in XAI within the project will pave the way for increased transparency and trust in AI systems within the manufacturing industry.