#It stands to reason that over sufficient repetitions videos made by this 'closed' system would standardize into a multimodal output model

[endreal]

brb, hooking an ai system trained to write scripts to an ai system trained to produce video from textual input and an ai system that generates descriptions of video content as a research exercise in identifying the most prevalent tropes and plot beats in modern cinema by (manually) cross-comparing discrete productions once content has stabilized at statistically significant similarity.

#I don't think it'd be easy but I do think it'd work. #Since modern 'ai' is basically programmatic pattern automation and amplification #It stands to reason that over sufficient repetitions videos made by this 'closed' system would standardize into a multimodal output model #Based around repeated trends extracted form initial input data #My hypothesis is that thru repetition trends would become refined + amplified to the point of being recognizably discrete and identifiable #My secondary hypothesis as that the system would generate video output unrecognizable as films by human standards #But *still based on ai interpretation of these trends/tropes* #Which is be fascinated to see the presentation of bc I want to know how one would have to engage with this wholly automated output #To re-interpret it into a human accessible media/analysis framework