This project uses artificial intelligence to classify artworks according to the Dutch concept of “gezelligheid,” a word associated with warmth, comfort, and pleasant atmosphere. I trained an AI model to evaluate artworks and determine whether they are “gezellig” or “ongezellig.” The results reveal the strange and often humorous logic of algorithmic interpretation. Classical paintings are suddenly judged through this cultural filter, sometimes producing unexpected or absurd outcomes. The project also reflects on the human influence behind AI systems. During training I repeatedly labelled images of asses as “gezellig,” which caused the system to strongly associate this category with them. This moment of feedback exposed my own subjective role in shaping the dataset and highlighted how algorithmic decisions often reflect the biases and choices of the people who train them.