- Share this article on Facebook
- Share this article on Twitter
- Share this article on Email
- Show additional share options
- Share this article on Print
- Share this article on Comment
- Share this article on Whatsapp
- Share this article on Linkedin
- Share this article on Reddit
- Share this article on Pinit
- Share this article on Tumblr
Is ChatGPT a sign that automation is coming to film and TV writing? As far-fetched as it sounds, the arrival in November 2022 of a free prototype of the AI-powered chatbot — which has jolted observers with the sophisticated, fluid writing it can produce when prompted, even in the form of poems, essays and, yes, short scripts — has set off alarm bells about the disruption that the chatbot could wreak on the work of entertainment scribes. Still, top film and TV writers are skeptical that the technology in its current state imperils their livelihoods in any way, even as they remain cautious about the potential for future advancements.
“Do I see this in the near term replacing the kind of writing that we’re doing in writers rooms every day? No, I don’t,” says Big Fish and Aladdin writer John August, who has tested the free research preview and talked about it on the popular Scriptnotes podcast, which he co-hosts with Craig Mazin (The Last of Us). Still, he adds, “There certainly is no putting the genie back in the bottle. It’s going to be here, and we need to be thinking about how to use it in ways that advance art and don’t limit us.”
Another prominent writer and showrunner, speaking to The Hollywood Reporter anonymously, has taken ChatGPT for several test rides and says the chatbot seems incapable of writing funny jokes or producing results that might be useful to include in a script without “substantial creative input from me.” This showrunner adds, “When people conclude that this is going to replace professional writers, I think they’re sort of swallowing an Elon Musk-style fantasy about the future that is not actually connected to the technology.”
Launched by the AI research company OpenAI, ChatGPT in its publicly available iteration can produce polished, if rote, pitches and loglines for films and television shows as well as generic outlines and scripts within seconds. The current version, trained on large quantities of text and code all predating the fourth quarter of 2021, also occasionally produces some falsehoods when answering factual queries: OpenAI CEO Sam Altman has warned that this version is “incredibly limited, but good enough at some things to create a misleading impression of greatness.” He added, “It’s a mistake to be relying on it for anything important right now. It’s a preview of progress.”
Still, several writers say the incipient technology shows promise as an ancillary creative tool. ChatGPT could help with time-consuming rote work for writers, like generating potential scene locations or character names; August suggests that non-native English speakers might use it to produce more fluid writing in the language. It can also offer plot or character ideas. Savage Grace writer and former WGA West president Howard A. Rodman compares the chatbot to “Oblique Strategies” cards that aim to re-inspire artists with prompts: He says he could use the bot “to suggest something I would never think of myself, maybe something smarter, maybe something stupider or more obvious.”
There is cause for concern, for some, about future refinements to this kind of technology. Rodman calls ChatGPT’s clear ability to improve its responses the more prompts it receives “a little chilling,” but adds, “Writers should not blind themselves to the ways in which AI technology can be useful in many ways. They should also be aware of the opportunities that it offers employers to do the thing they love best: put downward pressure on fixed costs.”
Franklin Leonard, founder and CEO of screenplay platform The Black List, doesn’t worry about ChatGPT replacing writers, but he believes it has the potential to upend the labor market and says it’s “necessary that the communities that are likely to be most affected by it, in particular writers, are the ones who are defining the guardrails around how it can be used in the context of this business.” Although he’s not privy to potential conversations, Franklin hopes that the Writers Guild is discussing this technology, as he believes the long-term financial health of the industry is “incredibly closely tied” to that of writers. (The Writers Guild of America West said in a statement, “We’re monitoring the development of ChatGPT and similar technologies in the event they require additional protections for writers.”)
Even as the technology is rapidly advancing, studios likely won’t roll the dice in exploiting works — scripts, art or otherwise — solely generated by AI just yet. There’s no copyright protection for such works because intellectual property law doesn’t recognize ownership by nonhuman creators. AI-created works would enter the public domain upon release, potentially limiting commercial interest in the format.
There’s been a push for protection of works created by AI spearheaded by Stephen Thaler, the CEO of neural network firm Imagination Engines. In 2018, he listed an AI system, the Creativity Machine, as the creator of an artwork called A Recent Entrance to Paradise, while listing himself as the owner of the copyright under the work-for-hire doctrine. The U.S. Copyright Office denied the registration, finding the work “lacks the human authorship necessary to support a copyright claim” and that “the nexus between the human mind and creative expression” is a crucial element of protection. Thaler sued the office in June over the rejection, but the office said it’s unwilling to “depart from a century of copyright jurisprudence.”
“The U.S. Copyright Office has looked at this topic through a very narrow lens,” says Joel Feldman, a lawyer whose specialty is copyright and trademark litigation. He stresses that the copyright office would be more willing to grant protection to AI-generated works as long as there’s a human co-author. “If you fed your favorite 25 songs into an AI machine, and it spit out a new work never before heard, there’s certainly a claim that your selection of those 25 songs is human authorship.”
Still, Ryan Abbott, Thaler’s lawyer and a partner at Brown Neri who specializes in the intersection of AI and copyright law, notes that the protection of works generated by AI is a question of “when” rather than “if.” “This is another instance of the law having to catch up with the state of technology,” he says. “When lawmakers were drafting copyright law, they weren’t thinking about AI.”
Whether AI programs, built on machine learning models that analyze the patterns of copyrighted works, infringe on the copyrights of artists is up in the air. Courts may find that training an AI program like OpenAI’s DALL-E 2, which generates complex images from text prompts, using copyrighted art does not fall under the fair use defense, which allows for use of protected works without permission as long as they are transformative. In November, a proposed class action was filed against Microsoft, Github and OpenAI claiming the billions of lines of computer code that their AI technology analyzes to generate its own code essentially constitute piracy. A finding of copyright infringement in the first-of-its-kind suit against AI programs that don’t acknowledge prior work would severely dampen the economic prospects of exploiting works generated by AI in Hollywood.
The prominent showrunner suggests that, if studios ever try to input their scripts into a database and use that to make a text generator in an attempt to bypass the work of a human writer, “they’re going to have tens of thousands of angry creatives on their hands.” He adds, “I’m not sure that they’re going to be interested in dynamiting all of those relationships in one fell swoop.”
Agents are also keeping a close eye on potential rights issues. CAA’s chief metaverse officer, Joanna Popper, says that the agency is endeavoring to work with cutting-edge tech companies (such as deepfake firm Deep Voodoo, from South Park’s Trey Parker and Matt Stone, which in December received $20 million in a funding round led by CAA’s Connect Ventures) to keep emerging tech talent-friendly. “I hope the industry will respond [to ChatGPT] by seeing that it’s a tool that could benefit them, while also being cognizant of making sure that their rights are honored,” she says.
AI-Driven Script Pitches, as Critiqued by a Human Screenwriter
THR asked ChatGPT to generate a series of feature film elevator pitches that mix and match established hits — a time-honored industry practice — then asked Big Fish scribe John August how well they worked.
Die Hard meets You’ve Got Mail
“When a group of terrorists take over a high-end Manhattan department store on Christmas Eve, a scrappy small-business owner must team up with a tough cop to take them down and save the day, all while trading banter and flirting over email.”
“‘Small-business owner’ doesn’t bring anything helpful here. If the Meg Ryan equivalent were an anti-cop activist, there’d at least be some tension.”
Fast and the Furious meets Independence Day
“Street Warriors: When a group of skilled street racers are recruited by a top-secret government agency to fight against an alien invasion, they must use their expertise behind the wheel to save the world from destruction. As they race against time, they must push their limits and embrace their inner strength to become true heroes.”
“The second sentence is meaningless filler. We need some sense of why they’re recruited, like the drillers in Armageddon or the hero of The Last Starfighter.”
The Hangover meets La La Land
“Lost in L.A.: When a group of friends travel to Los Angeles for a wild bachelor party, they wake up to find that one of their own has gone missing. As they search the city for their missing friend, they are drawn into the vibrant world of L.A.’s music scene and must navigate their way through a series of zany mishaps and absurd adventures.”
“Weakest of the three. It’s literally just The Hangover. A true mashup might have boy and girl searching for each other after a wild night.”
John’s Overall Takeaway:
“These read like the summaries on the top sheet of script coverage. That’s not a knock; I used to be a reader at TriStar, and they’re not always easy to write. As far as the ideas themselves, they’re all taking ‘X meets Y’ far too literally. Screenwriters don’t start with this shorthand; they find their way to it to explain their take.”
A version of this story first appeared in the Jan. 11 issue of The Hollywood Reporter magazine. Click here to subscribe.
Sign up for THR news straight to your inbox every day