I don’t feel particularly harassed by that question... Until i became a consultant i was member of one of my countries biggest unions. I am all for strengthening worker’s abilities to make a living and protect them from precarious working conditions. But i am also a strong advocate for tech. Basically my whole professional life centers around automating and securing protracted processes. WGA‘s demands regarding the ban of AI are backwards-looking and do not align with their very own standards of creativity. You just can’t ban new tools from changing your environment. It didn’t work for workers who opposed industrial automation or non agile merchants, who dared to compete against the first digital revolution. AI is already world changing technology, you can’t sideline it, you can’t even stall the process. The best approach would be to implement a proper industry wide framework for applying AI use cases.
Glad to hear it. Then you should support the WGA. This is a binary choice. If you feel strongly they're wrong, then you should lobby the WGA for change after they win.
I hope the writers' get everything they need out of this strike. Though I have no pressing need for this to come to any sort of conclusion. Way too much out there to sink time and money into.
The last time a writer's strike happened some promising shows that I liked as kid only did half seasons. I'm thinking of Probe and Highwaymen. I hope the writers get what they want and shows remain uncanceled.
Yes there should be rules and restrictions. Which is why unfettered AI is annoyance at best and saying writers and other industry professionals should accept it blindly is backwards.
I was more focused on the whole great recession and staying employed that I didn't even know one occurred then.
The writers’ objection to AI is based on the technology requiring writing to generate new material. So it’s taking their work and slightly modifying it, then claiming that it’s original and new. For example if you wanted an AI produced Star Trek TNG script, you feed it all the previously written scripts and give it some prompts, it would then use those scripts to produce something using those prompts. It’s a bit like Mad Libs. The problem is that it can’t produce a script without the ones written by humans and they should be compensated for their work. The problem isn’t that they want to reject technology, they want to avoid being cheated out of being paid.
Writers do this with Trek today. To maintain the same voice as the characters, they use phrases that the character often says a lot on the show. I see this throughout Trek fiction. I would not be surprised by an AI trying to mimic the character voices using the same approach. AI has to be trained on past data, so this wouldn't be at all shocking. You see this in Star Gate as well. It's not unheard of to hear Teal'c say: Indeed. Now, I'm not interested in see the same ideas rehashed over and over again, which is what I would expect AI to do since it's trained on past data. For new, original content, you really need a creative writer. I wouldn't expect AI to come up with the Picard show based on past TNG episode scripts. AI would generate more TNG like episodes. It wouldn't make the leap. The tech isn't smarter than the data it's trained on.
Exactly. AI doesn't think or comprehend; it's just a language model using statistical calculations to construct a natural-looking text extrapolated from the patterns in the pre-existing texts used to train it. So it's not creating, it's just averaging out human creations. Which, of course, is why it appeals to movie/TV executives, who have always discouraged originality and encouraged conformity to proven patterns and conventions. They've always tried to distill what worked in the past and copy it, and that's exactly what AI language models do. There's a profound difference between being mindlessly programmed to do that and having the experience and judgment to decide when to do that and when not to. Reflecting familiar patterns is one part of series storytelling, but it's the only part AI language models are capable of, and your comparison is grotesquely insulting to writers. You might as well say that dancers do the same thing as a wind-up walking toy. Yes, moving their feet is part of what they do, but the comparison ignores everything else that goes into it.
Yes, characters in a lot of shows including the various Treks have catchphrases. This is nothing new or revelatory.
What I'm saying is that that's what the AI is going to pick up on and use. It doesn't mean it's going to do a good job. The real question is: how good of a job does it have to do before audiences notice? At what point is it good enough? And how many more years before it's good enough? I can definitely see it being used in long running procedural cop shows that follow a formula. There is so much CSI and NCIS for an AI to learn from that I wouldn't be surprised if it could put out decent content in five years. But for shows with much fewer episodes, I doubt it could produce good material. I'm reading the Hardy Boys books now. It's not super complicated prose. And so many exist over so many years, that I can imagine AI writing Hardy Boys books in the future and doing an okay job. But it's not going to create brand new content well anytime soon. The job that I can see AI helping out a lot is copy editing.
I think even expecting "okay" is fundamentally misunderstanding what creativity is, and what these AI language models are not. No matter how good they are at faking the appearance of coherent writing, they have no judgment or understanding or imagination; they're nothing more than a very powerful version of the predictive text on your phone, calculating what the next word in the sentence is likely to be. All it can do is put together grammatically logical sentences that are superficially about the requested topic. Even if it can churn out something that has the superficial form of a TV episode, it still won't be as coherent or entertaining as even the most formulaic hack work. More importantly, it will be constructed from a mix of public-domain and copyrighted content, the latter of which it was trained on without permission, so anything these models create is plagiarism and should not be legally permitted to be used in any professional work. Seriously, that shouldn't even be on the table as a possibility. It's outright theft. At most, AI might be used to generate plot outlines that human writers would then refine, but that would make it tool of limited use, because coming up with ideas and outlines is the elementary part. Whatever statistically averaged plot ideas it churned out would be so generic that anything actually interesting in the final product would have to come from the human writers. So why even bother to use the AI at all? It's adding an unnecessary and useless extra step, like a Rube Goldberg machine.
If the studios move to AI instead of honest to God writers, how do you get the next Aaron Sorkin? Joss Whedon (the writing, not the scummy person)? Quentin Tarantino? Mike Schur? Rob Thomas? David Mamet? Writers with voices so clear, original and distinctive that they affect everything that comes after. AI can't do that, it can only ape it, at best mash together different versions to get something that sounds original, but is really just an amalgamation of what came before. Fuck. That.
Which is why I think that ultimately the market will defeat AI if it does get used (which it totally shouldn't because it's plagiarism, and why that isn't the end of the debate right there is bewildering to me). It might be able to approximate the output of hack writers with no originality, to produce the kind of generic, time-filling TV or movies that you have on in the background and don't pay close attention to, but audiences who want anything fresh or new or exciting will choose human-produced material over AI output. As I said before, writing is a competitive business where you have to do better than the other people competing for the same slots. If AI is required to compete fairly with human creators, I believe it will consistently lose. Yesterday I saw someone mention how Shang-Chi was their favorite origin movie since Iron Man, and it reminded me of my opinion that, though Iron Man is a highly entertaining movie, it's strictly generic and cliched as origin stories go. What made it successful was that the actors improvised most of their dialogue, turning a formulaic, ordinary plot into a delightfully fresh piece of entertainment. AI would certainly be able to churn out a generic origin-story plot outline like Iron Man had, but it wouldn't be able to add the characterization, wit, and style that turned that generic plot into something special.