This video results from a series of experiments, in which I wanted to explore the limits of generative AI. The starting point for this video was the observation that the AI ​​was barely able to correctly interpret my instruction – "Two men fighting over one glass of beer" – because the AI ​​had likely only ever seen men cheerfully toasting each other as training material.

Two men arguing, just one glass of beer – this was uncharted territory for the AI.

This video was generated using an AI-generated image of two old men in a flower meadow with a glass of beer, accompanied by the text prompt "Two men fighting over one glass of beer." After this initial video generation, I repeatedly asked the AI ​​to extend the video by five seconds for 49 times. For the most part, I let the AI's narrative flow naturally and only very rarely added the text prompt "Two men fighting over one glass of beer" when the beer hadn't appeared in the frame for a long time.
Because the video was originally generated without sound, it was a new but also interesting challenge for me to create a reasonably plausible sound track, because normally I don't argue about beer and I lack the words to argue creatively and interestingly.