Artificial intelligence (AI) is popular for applications in image or natural-language processing, but AI can also be used to learn complex relations in production processes. For example, an AI can predict product quality based on process data during the production. In this paper, we present an application of AI to estimate structural properties of additive manufactured components in real-time. Occurring imperfections, such as air inclusions in the component, are considered and evaluated, since these have a significant influence on the quality of the component. This approach combines finite element (FE) simulation and machine learning: based on FE simulations, a neural network is trained to represent the relation between imperfections and the robustness of the component. To predict the impact of imperfection in real-time, monitoring systems are used to detect anomalies during the printing process, which are indications for imperfections in the additive manufactured component. Afterwards, the trained model is used to evaluate the impact of the detected anomalies to the component quality. This application of AI has a great potential to improve the additive manufacturing process itself and simplifying the approval of additively manufactured components.