Creativity as a process, possibilities and limits of AI in Films and Moving Images

What does it mean to govern AI in creative production if creativity is understood not just as a finished result, but as a process? A complex web of experimentation, judgement, and revision, all filtered through human emotion and experience. Drawing on film and moving-image practice, this talk examines what we usually define as the creative process.

The talk explores the probabilistic mechanism of generative AI, which produces outputs based on patterns learned from existing material. These systems can widen access to visual experimentation and also help filmmakers work within time, budget and technical constraints, but their outputs consist of statistically plausible results rather than intentional creative expression. The AI models also reflect the cultural assumptions, hierarchies and biases present in their training data. The question is not simply what AI can create, but which aesthetic conventions it reinforces, which perspectives it marginalises and whose work enables the output.

The adoption of AI-generated visual media raises further questions about taste and culture as their aesthetic patterns may influence what audiences recognise as desirable, original or realistic, widening the gap between deliberate creativity and statistical prediction.

Questions extend beyond copyright, consent and transparency. Who determines the visual norms these systems reproduce? Who benefits from their adoption, and who bears its costs? How can governance protect creative freedom without restricting experimentation?

Governing AI in the moving-image industries therefore requires more than ensuring that outputs are lawful. It means establishing clear rules for the use of creative work and performers’ likenesses, improving the traceability of generated images, protecting filmmakers’ creative control and assessing the effects on employment and visual diversity. It also means asking who sets the standards of visual culture as AI becomes widespread.