Additive manufacturing allows components to be produced in an optimized manner that is appropriate for the load path and material. This requires in-depth knowledge of the relationships between the processes involved, the component structures and the component properties, which can be incorporated into the design process of the components. If this knowledge is not available, the components can fail prematurely under load. For a systematic generation of knowledge, data of the processes involved, the materials used and the test procedures carried out on additively manufactured components made of Ti6Al4V are recorded in the AMTwin research project. The entire dataset contains key figures for each process, including parameters characterizing the microstructures determined by numerical simulations. The recording of the data is part of a systematic research data management, which is absolutely necessary for the functioning of the joint project AMTwin. In this contribution, the quality of additively manufactured components is predicted using AI-based methods based on the recorded data set. For a better understanding and characterization of the data set, explorative data analysis methods are used first, including parallel coordinate plots, which enable an easily understandable visualization of process-structure-property linkages. Typically, small batch sizes are used in additive manufacturing, resulting in data sets with small sample sizes and many different parameter combinations. The data set used here contains key figures of around 1000 additively manufactured components. Therefore, different AI-based approaches are proposed and applied that take these boundary conditions into account. The results are discussed, the main influencing factors are shown and conclusions are drawn for the optimization of additively manufactured components. The presented results are part of the research project AMTwin, which is funded by the Free State of Saxony.