In the area of microscopy, machine learning methods are increasingly used to evaluate images—in areas such as chemistry, pharmaceuticals, medicine, and material analysis. After they have been “taught” once, algorithms are superior to human experts, especially when it comes to segmentation; in other words, filtering out relevant image information. This is particularly true when a huge number of images or large areas have to be analysed. Neural networks are trained to identify patterns at top speed: Material defects, such as inclusions or micro-cracks, as well as the position and contours of specific cells, cell nuclei, or other cell components. This is often possible based alone on the grey-level values and shapes so that no fluorescent markers have to be used. As a result, there is no need for time-consuming sample preparation, and fluorescence channels remain free for other markers. But there are further benefits. Machine learning helps validate the information of a sample with various microscopy methods. Once it has been taught accordingly, artificial intelligence can also be used to discover extremely complex cause and effect relationships by inferring the material behavior from structural analyses—and, as a result, discovering previously hidden regularities.