“This machine can produce a 5,000-word story, all typed and ready for despatch, in 30 seconds. How can the writers compete with that?” So asks Roald Dahl in his short story The Great Automatic Grammatizator, published in 1953.
The eponymous machine learns from existing works and can write entire novels in response to manual prompts. By pressing buttons and pulling stops similar to that of an organ, the machine’s inventors “pre-select literally any type of plot and any style of writing”. They choose between “historical, satirical, philosophical, political, romantic, erotic, humorous, or straight” categories, then between themes such as childhood, civil war, or country life, and finally between literary styles such as “classical, whimsical, racy”,or specific authors such as Hemingway, Faulkner and Joyce. Within a year, half of new writing is generated by the machine. The ending of Dahl’s tale is harrowing.
More than half a century later, the fear felt by some writers, artists, and musicians, when faced with generative artificial intelligence such as ChatGPT, is not unfounded or foolish. Those who dismiss it as posing no threat to the creative industries, on the grounds that it can never be good enough, are the deluded ones; AI is more than capable of producing content comparable to that created directly by humans.
Contrary to popular opinion, however, this is not intrinsically dangerous. Rather, the danger is us — or, more specifically, our refusal to engage with it seriously. The current state of AI in the film industry, and the varied reactions to it, exemplifies this. Already, generative AI’s role in film ranges from writing scripts to creating visual imagery, and its adoption has sparked a range of reactions including celebration and anger. Few and far between are those who say simply and rationally: it’s here, so let’s see what it can do, and let’s talk about it, because in doing so, we can shape our creative future.
In this camp sits Swiss director and producer Peter Luisi. When ChatGPT 4.0 was released, Luisi wondered whether it could write an entire feature film. He set about finding out, ensuring his input was that of a typical director, and was not enough to grant him a writing credit. ChatGPT 4.0 was given the initial prompt to “write a plot to a feature-length film where a screenwriter realises he is less good than artificial intelligence in writing”, and asked to generate each scene in turn. Screenshots of the prompting process are available on the project’s website along with the film itself: The Last Screenwriter.
London’s Prince Charles Cinema was hired for the premiere in June, and in keeping with the aim of stimulating discussion, the team made tickets publicly available (although the film is not-for-profit, they decided to charge admission to discourage no-shows) and agreed to pay for promotion. But when posts went up on the cinema’s Twitter and Instagram, there was a furore; many were angry that the cinema was apparently supporting the use of AI in film. The Prince Charles cancelled the screening, citing “the strong concern held by many of our audience on the use of AI in place of a writer”.
When I spoke to Luisi the next day, he was sanguine — “We were sort of hoping that this discussion would happen, we were just hoping it would happen inside the cinema” — and invited me to the relocated premiere. Sitting there felt utterly surreal, a new page in a history book. How good ChatGPT’s screenplay was didn’t seem to matter: what was significant was that we were watching the first film written by a machine. This was just the start.
Join the discussion
Join like minded readers that support our journalism by becoming a paid subscriber
To join the discussion in the comments, become a paid subscriber.
Join like minded readers that support our journalism, read unlimited articles and enjoy other subscriber-only benefits.
Subscribe