With an increasing number of firms now actively working on Multimodal Models and Novel Hardware, not only in unique art form but also in underlying technology, with an added compound factor of regular business competition driving an Evolutionary push toward more cost effective while more potent models, adding breakthroughs such as character and scene continuity, physics mimicry, increased output lengths, scene-weaving, addition/subtraction from output, text-to-voice, voice-to-video, text-to-specialeffects, output editing, as the prohibitive cost decreases, the number of net profit independent producers increases, which further increases the number of attempting producers.

Modifiable outputs in LLM storywriting allows for customizable storycrafting, which can then be polished and crafted into non-slop, to which the storyboards can be passed though. With increased complexity reduces the number of personhours spent refining the outputs. As context windows increase dramatically with fractions of a cent per 1 thousand tokens for even today's most powerful models, Eventually, human intervention reduces to zero with a fine enough tuned data sets.

One could even go as far as to pipeline multiple models from entering in a descript and complex prompt, the LLM writes a 300 page, concise, engaging and plot-holeless story in ANY genre, then automatically feed a page at a time into a Multimodal model of ones choice, generating an entire AAA feature length film in 24 hours.

Several conspiracy theories place this pipeline as a potential reality in the music, TV and film industry for decades, soon, ANYONE can cook.

TL;DR: When the poopoo lines hit the peepee lines, you're going to see some serious shit.