Manufacturing quality is undergoing a significant shift as Industry 4.0 technologies introduce new ways to monitor, predict, and prevent defects directly in the production process. Classical methods such as Statistical Process Control or Six Sigma address manufacturing issues at a later stage, while current AI-based solutions still struggle with limited contextual data, high variability, and low trust from operators, especially in SMEs. This research explores how AI can be optimized for human-centered manufacturing quality by combining real-time sensor data with operator feedback to create adaptive learning loops. Algorithms are continuously updated through reliability scores and transfer learning, ensuring that the system evolves with the production environment and remains aligned with operator insights. In doing so, the approach integrates process- and product-related perspectives while making advanced quality monitoring more practical and accessible on the shop floor.