AI art refers to digital works created using artificial intelligence. However, the term is used broadly rather than as a single fixed category. What is commonly called “generative AI” today is described as an output system based on machine learning, especially large deep-learning models trained on massive datasets. Such systems optimize outputs according to given data, objectives, and evaluation criteria, and their internal representations and decision-making processes are often difficult for humans to interpret, making them effectively black boxes.
AI art is also approached from a critical and conceptual standpoint. In some practices, AI is not treated merely as an automatic image generator, but as a conceptual medium for abstracting and visualizing an artist’s ideas. In legal and institutional debates, human agency remains important: the construction of datasets, customization of algorithms, deliberate disruption of models, and selection of final outputs are all emphasized as part of the creative process. AI is also distinguished from ALife: AI-based generation is described as more top-down, while ALife systems such as cellular automata generate changing visual and sonic phenomena through repeated local rules and bottom-up interaction.