Charlie Brooker’s cult anthology series is back! The sixth season of black-mirror will be available this Thursday on Netflix. With the thunderous arrival of the chatbot ChatGPT, Charlie Brooker, the creator of the famous dystopian series, couldn’t resist the urge to ask the AI to write an episode of black-mirror. “I played a little with ChatGPT”, he says in the new issue of the British magazine Empire. “The first thing I did was type ‘generate an episode of black-mirror“And it gives something that, at first glance, reads plausibly, but at second glance, it’s crap,” he believes. The rapid development of new generative AI tools as ChatGPT is therefore currently at the heart of the concerns of the various audiovisual professions, so what role can AI play in the creation of series?
American screenwriters from the Writers Guild of America (WGA) have been on strike since May 2. At the heart of their demands, two questions, that of the remuneration of creation, but also that of their profession threatened by artificial intelligence. The WGA wants to “regulate the use of material produced using AI,” the union says on Twitter. Artificial intelligence “cannot write or rewrite script material”, the union believes, and therefore cannot be considered an author.
“There’s not really any original thinking here”
These are exactly the conclusions of Charlie Brooker’s experiment: “Because all ChatGPT does is search for all the synopses of episodes of black-mirror and mix them. Digging in, you think, “oh, there’s not really any original thinking here. The screenwriters refuse to recognize any special status for material originating from the AI, intending to retain all of their copyrights.
Same story with Lisa Joy, screenwriter, producer, director and showrunner, to whom we owe the series Westworld. “ChatGPT is an information aggregator. It collects what already exists on the web. ChatGPT don’t think so. It assimilates, finds tendencies, digests them and spits them out. It does nothing new in itself, ”she said during a master class at Series Mania.
“The idea comes from the human imagination”
“AIs will become tools that will help us in creation”, she had developed during an interview with 20 minutes at the Lille festival. And to recall: “They have already been there for some time. The trailers are already made with illustrations generated by AI. But this AI doesn’t come with the idea of what the trailer should tell. He simply executes, the idea comes from the human imagination. »
A feeling shared across the Atlantic. “Writing is difficult to reproduce. With the team ofHere it all starts, we had fun asking ChatGPT to write a scene, so of course, these are the beginnings and we can imagine that it will improve, but the result is an empty shell. The human being, which is at the heart of our profession, cannot be imitated”, considers Solenn Le Priol, one of the scriptwriters of the daily soap opera of TF1.
The American directors, who have joined the screenwriters in their demands, have found an agreement that holds in this formula: “IA is not a person” (“AI is not a person”). “In any form of post-production or visual effects, we already use the help of computer technology. It’s not AI yet, it’s a little more rudimentary than that, it’s just a tool, ”sums up Lisa Joy.
“Who owns the job? Who is the creator? »
Tools that offer many possibilities that worry American players, who began negotiations on June 7, via SAG-AFTRA, their main union across the Atlantic (one hundred and sixty thousand members). The latter want to impose new clauses on AIs in their contracts. The idea is to protect against the possibilities of digital rejuvenation (as in the next IndianaJones expected in theaters on June 28), or even the appearance of a deceased actor (like Peter Cushing in the saga Star Wars) or simply being replaced by virtual actors.
The use of AI thus raises many questions, but does not prevent us from considering its use in the creative process. “Who owns the job? Who is the creator? But also ethical questions, about copyright, as well as deeper philosophical questions about the future of humans compared to machines,” considers Merav Shacham, who oversaw the creation of the credits for the Israeli series. Red Skies, presented in official competition at Series Mania and produced using an AI.
“90% of what I produced, I had to throw it away”
“I started researching AI image generators five or six months ago. I experimented with multiple platforms and created dozens of animation videos based on artificial intelligence,” she explains to 20 minutes. When Ron Leshem, writer and producer of Red Skiesoffers him to make the credits of his series, she offers him to do it using the Stable Diffusion image generator and “to see what was going to happen”.
Merav Shacham, Ron Leshem and director Alon Zingman are writing a specially adapted screenplay. The main difficulty? “We had to learn to communicate with the AI so that the result obtained was as close as possible to the written scene. I had to learn how to write prompts. The AI is very difficult to control, 90% of what I produced, I had to throw it away because it was not good”, she says. Especially since the AI was regularly updated. “So I had to adapt quickly to these daily changes,” she continues. The process is a “real challenge”, not least because the visual productions of the AI ”have a disturbing, often surreal and incoherent aesthetic”.
Result of this experience? “Each artistic project has its own nature and its own needs. AI, at least today, is not appropriate for all projects. I used it as a tool, not as something that will replace me. We must learn to use it intelligently. this tool can serve me as a creator thanks to a fruitful collaboration with him and an in-depth dialogue”, analyzes Merav Shacham.
“The myth of AI thinking for itself is damaging”
“Humans are also there to shape these tools, and to define their powers and limitations. The myth of the AI which thinks for itself is damaging”, abounds Lisa Joy. Behind every AI, there is human production. “They are not yet forces of creation, but we can perceive them like that. If people start acting like they are, like independent spirits, then we’re going to have trouble,” she continues.
“As a screenwriter, we use the Internet a lot for research. The problem is that the AI talks a lot of nonsense. This will cause problems, ”defends John Kare Raake, the creator of the Norwegian series TheFortress. AIs like ChatGPT “only reflect data available on the Internet. They collect this data as facts, but that doesn’t mean it’s true. There are a lot of racist remarks on the Web for example. AIs have no moral compass,” adds Lisa Joy.
These conversational agents will still perhaps come to the aid of the scriptwriters. “Sometimes I pray that it can come to save us from situations where I am blocked”, laughs Cécile Berger, one of the screenwriters of the Red bracelets, new generation. “AI might be useful when you’re stuck having someone to talk to… It’s a tool in the creative process,” added Cecilie Mosli, one of the directors of Fortress.
There will always be fears about any new technology out there, as Charlie Brooker demonstrates in black-mirror. “These tools are there, they exist and nothing can be changed. It is fascinating and interesting. But it is not because a technology appears that it will change everything. It’s going to be used and modeled. We will see a lot of things done with the help of AI in the next few years, ”concludes Jonas Weydemann, co-creator of the German series. A Thin Line. Series created with the help of AI, but born in the human imagination for a very long time!