Human creativity and generative neural networks: similarities, differences and perspectives

Generative models are tools capable of both exercising highly sophisticated combinatorial artistry and grasping the space of possibilities based on what they have "seen." If the world corresponds to the dataset on which artificial intelligence is trained, albeit limited compared to the "dataset" of the human mind, it potentially contains all future artistic productions, and the machine is not only able to capture the sphere of these possibilities but also to exploit them in ways often novel if compared to human statistical norms. Therefore, the recent developments in AI in the artistic field, coupled with the increasing pervasiveness of artificial generative networks, seem to challenge the notion of creativity as an exclusive property of human beings. This calls for critical reflection on human artistic creativity and its presumed artificial counterpart.