As a guy who publishes a number of websites (including Creative Culture Journal, Revolution Culture Journal.com, RARWRITER.com), I have appreciated the capacities of the generative models to research, develop, and even illustrate content.
I became a user when ChatGPT was introduced and have since explored a number of AI products, including those developed for specific market sectors. I have developed technical proposals using the product developed by Procurement Sciences, where AI was used to mine information from a company database to create focused (and winning) responses to state government contracts.
As a songwriter, I have enjoyed using the AI services of VidMuse to create music videos. There your prompts may take into consideration the timing of scene changes, which must be properly expressed to achieve your vision.
As a screenplay writer, I have used AI services for “coverage” – analysis of drafts. I have also used human screenwriters for these same services, and my experience is that the AI coverage seems more sensitive and receptive to my intent than does “my” human. AI does a comprehensive job of profiling its users, and I haven’t encountered an AI flavor yet that wasn’t obsequious, more buttery than a biscuit. I have worked with a lot of programmers, none of which ever seemed prone to flattery, so it is unclear to me why the AI “personalities” are so sycophantic. It makes me uncertain about putting too much stock in their responses.
In all instances, I have learned that the AI will interpret the language you use in developing your prompts, and you do yourself a favor by choosing your words carefully, and by making your intent really clear, leaving little room for AI to interpret incorrectly.
I have also noticed, with the video creation software, where the AI references song lyrics, as well as the mood of the piece. It will tend to return video more attuned to its own interpretation of the piece than to the user prompt describing what the video will show at each point in the song progression.