Đorđe Lazarević
Faculty of Philosophy at the University of Niš, Serbia
Recent advances in “creative technology”, particularly AI-generated images, poetry, and scientific ideas, have raised questions about whether machines might one day surpass human creative abilities, or whether they already have. To properly assess these and related questions, it is crucial to first establish a clear conceptual framework for creativity itself. In this talk, I pursue two main objectives. First, I propose an analysis of creativity by distinguishing three key components – product, process, and agent – and offer a working ontology of creativity that clarifies the relationships among these elements. Second, I briefly explore the relevance of agential and productive aspects for AI, before turning to a more pressing question: Can a machine be spontaneous? Since spontaneity is a necessary condition for any genuinely creative cognitive process, even if a machine were to have a conscious desire to create, its presumed lack of spontaneity would imply a lack of creativity.
One common argument holds that machines lack spontaneity because they invariably follow instructions and therefore lack intrinsic motivation and creative freedom. I challenge this argument in two ways. First, both thought experiments and empirical studies show that human creative actions can also be determined and predicted, which undermines the idea that spontaneity requires indeterminacy or unpredictability. A stronger version of the argument, however, claims that AI agents are always causally dependent on human input, whereas human beings, at least in some cases, initiate creative processes independently. Here, I suggest that the key distinction lies in how we interpret human versus machine action: we attribute creativity to humans partly because we lack full knowledge of the causal factors underlying our own creative cognition, whereas we assume a fully determined model of action for machines. A further implication of this view is that a superintelligent AI—which would likely possess complete knowledge of its own creative cognitive architecture—could not be creative, precisely because it would lack the epistemic opacity that underlies spontaneity in human creativity, even if such an entity were capable of producing highly valuable ideas and artifacts.
Chair: Brian Ortmann
Time: 04 September, 11:20-11:50 (Cancelled)
Location: HS E.002
SOPhiA 2026 