Ruin Explorer
Legend
Dude. People do rewrites on the day. There may not be "major script revisions", but there could be fairly significant tweaks etc. - virtually every show does minor rewrites on a near-daily basis and those will all be lost. The dialogue will be clunkier, the scenes will work less well. It's just a question of how much those 19 days will matter - like how much they were actually going to get done?No major script revisions were coming at 19 days,this is symbolic (which is important I'm a strike, mind you!).
And they have lost the showrunners, which is pretty major. Any show that's consistent is going to feature regularly daily input from the showrunner(s). Only anthology-type shows can really dispose of them for a couple of weeks and say "that's fine!".
This is a pretty random damage roll on a large HP pool, I admit, but there's going to be damage, and it's certainly not purely "symbolic".
LOL.What will be interesting if this pushes studios to take the plunge and use chat gpt more for writing work.
This has been the precarious balance of worker negotiation. You can get better conditions, but push too hard and it makes it economical to automate you out.
Spoken like someone who either:
A) Hasn't seriously set ChatGPT to a writing task.
or
B) Doesn't understand what even mediocre writing looks like.
ChatGPT has a gigantic problem. It doesn't understand context or meaning, it only understands word order. That means it can only operate in cliches and tropes, essentially. If all you need is cliches and tropes, that can be fine. For example, if you want to use ChatGPT to come up with the background for an NPC who you don't need to be original and which the PCs won't read anyway, which will just inform you about them. ChatGPT is solid for that sort of task. It tends to overegg the pudding even if you try to ask it for restraint, and it tends to go for the most extreme cliches, but it's basically.
But writing dialogue? Hahahahahahahahahaha no. Absolutely not. Its dialogue is absolute total drivel that's only merit is the good grammar and punctuation. I've used it extensively, as have my friends, because we were fascinated. My wife who is both a writer and software developer has too. She's able to get good stuff out of it where good stuff is possible, but good dialogue, especially lengthy/complex dialogue, or dialogue with meaning/emotion, or dialogue that involves understand the relationship between two or more people? It literally cannot reliably do that. It can barely do it by luck. And whatever it gives you will be absolutely ridden with cliches and clunky/obvious dialogue, because it's been trained on a lot of extremely bad writing as well as good.
But again the key problem is it doesn't understand context.
You want an example? I got it to write a made-up outline of a Law & Order episode. What did it get right? The way the show is broken up into usually quite short scenes. The names of the detectives. The starting with a body being found. But it only lasted a few scenes before it derailed completely due to not understanding context - it had Benson and Stabler literally cleaning a crime scene themselves. Why? Because scenes where involving crime scene cleaners do happen in some L&O episodes - but it didn't understand, fundamentally, that a detective isn't the same thing as crime scene cleaners, because all ChatGPT is, is like nuclear-powered, ultra-large-dataset predictive text. It then went on to just entirely skip the court scenes, just focusing on the detectives. Again because it didn't understand context, it doesn't understand relationships. It just puts words one in front of another.
All it can do is spit cliches.
I will say, 50 or 60 years ago, that might have been "good enough" for an awful lot of TV - especially 1970s comedy.
That hasn't been "good enough" for even average TV (outside of truly dire crap like NCIS and the other absolute bottom-of-barrel lowest-common-denominator brain-off crime procedurals) for 30+ years. Even before the golden age of TV (which we are still in), even the 1990s, we were beyond that.
It also doesn't "do" continuity. Like it can't remember stuff. So you have to re-explain and remind it of all the facts every time, which means it's entirely unsuitable for anything long-form.