AI Portraits

By creating portraits of people using generative artificial intelligence trained on portraits of real people, I try to test my theory: 1. The person in the portrait himself is more likely not to recognize himself in the picture and will have disappointed or negative feelings. 2. The closest relatives will guess who is in the picture, but they will say that the portrait is not similar. 3. More distant friends are more likely to guess whose portrait is generated. 

Measurement errors show up due to special extraordinary features of the face. These errors are doubled by both the AI and the viewer. It is easier for both to generate and identify a model with characteristic features. Therefore, it would be better to test my theory on models with normal and ordinary faces. However, it is interesting to compare the results of generative portraits of people with diverse facial features.