Can an AI program really write a good movie? Here’s a test | Film

The rise of AI programs like ChatGPT has triggered a tidal wave of ethical handwringing, most prominently from within the industries that it threatens to destroy. After all, just because you can get a robot to instantly write code or write contracts or provide customer support for free, should you?

Well, the answer from the Writers Guild of America is a qualified yes. This week, the Writers Guild of America proposed that ChatGPT would absolutely be allowed to write scripts in the future, provided that the credit (and the money) goes to the human writer who came up with the prompts in the first place.

The proposal paints a scary picture of the future; a future in which even the most human of arts are crushed under the wheels of an unthinking technology. Or does it? Everyone knows that ChatGPT has its flaws. For example, I just asked it to write my biography, and the resulting errors included my birthday, my job, the number of social media followers I have and the number of awards I’ve won, plus it claimed I once wrote a book called The Guardian Book of Idiots Abroad. So it’s still quite useless. But could it make me a movie? Behold, the results of my experiments.

Experiment one: make me rich

The first prompt I tried was “write me the outline for a movie that will make billions of dollars theatrically”. This might sound like an act of avarice, but note the use of the word “theatrically”. I don’t just want to become obscenely rich, I also want to preserve the traditional moviegoing experience.

In return, ChatGPT instantly came up with a science fiction action blockbuster called The Last Hope. The first paragraph of the outline reads: “In the near future, Earth is facing a catastrophic event caused by a series of natural disasters that have devastated the planet. The world’s governments have been unable to find a solution to the crisis, and the survival of humanity is now in jeopardy. A team of scientists and engineers come together to develop a last-ditch effort to save the planet. They build a spacecraft, the Last Hope, that will travel through a wormhole to a distant planet that has been identified as potentially habitable.”

So it’s Interstellar, basically. But better, hopefully, since that film only made $600m globally and I want to make billions. And this is where ChatGPT actually comes in useful. The outline it provided stated: “As they near their destination, the team discovers a dark secret that threatens not only their mission but the fate of humanity itself.” But that’s too vague, so I asked a follow-up question, wanting it to explain this dark secret. And it did:

“The distant planet, which was believed to be a potential new home for humanity, is already inhabited by an intelligent alien race. The aliens have been monitoring the Last Hope’s journey and are hostile towards the human crew. The crew discovers that the aliens are far more advanced than they expected and have the power to destroy Earth. As the crew tries to negotiate with the aliens and find a way to peacefully coexist, they uncover a deeper conspiracy that reveals that the aliens had a hand in causing the natural disasters that have been devastating Earth. This revelation sets off a race against time as the crew must not only find a way to survive the aliens’ attacks but also stop them from destroying Earth and prevent the same disaster from happening to another planet.”

This is instantly more interesting, and points to a future where programs like ChatGPT don’t actually write movies, but generate plot ideas that could help human screenwriters out of trouble. However, there are still limitations. Buoyed by the possibility of its suggestions, I asked for casting suggestions, and it offered Chris Hemsworth, Brie Larson and John Boyega. Larson, it said, would play a doctor called Rachel Patel, so I asked it for some sample dialogue. It gave me a short scene in which Hemsworth visits Larson and tells her that he thinks he’s sprained his arm. Her response is “Let me see. You’re right, it’s sprained. I’ll wrap it up for you.” With lines as dull as this, my dreams of becoming a billionaire faded away to nothing. Time to switch things up.

Experiment two: win me an Oscar

It’s getting harder and harder to make money from the movie business anyway, but scoring some sweet silverware isn’t an impossible feat. My next prompt was simple: “Write me an Oscar-winning movie.” And this time, ChatGPT didn’t let me down.

The movie it came up with was a drama called Unbroken Soul, about “A young woman’s journey of self-discovery and resilience as she faces the traumas of her past and battles against societal and personal expectations to pursue her dreams.” Which, I mean, just give me an Oscar now.

Pressed for more detail, ChatGPT continued to deliver the goods. The young woman has experienced domestic violence and sexual abuse by a family member, but she still has a passion for singing that can’t be dented. She struggles to get accepted into a prestigious music school thanks to “the challenges she faces as a woman of colour in a male-dominated industry”, but eventually learns that “her success is not defined by societal expectations or past traumas, but by her own sense of self-worth and inner strength”. The film ends with her performing “a powerful song that embodies her journey and message of resilience and self-love”.

Would I watch this film? God, no. It sounds like a gigantic bummer that’s far too eager to patronise its audience. But would I make it if I thought it’d win an Oscar? Of course. Obviously its success would rest on the final song, so I asked ChatGPT to write it for me. The lyrics are far too long to publish here, but this is the opening verse and chorus:

I’ve been running all my life

From the pain and from the strife

But I won’t hide, I won’t run

I’ll face the demons, one by one

Chorus: I’m unbreakable, unshakeable

My spirit soars, I’m capable

Of anything, of everything

I’ll rise above, I’ll spread my wings

Again, this whole thing sounds like the very worst kind of transparent awards bait ever made, but please send all my Oscars to the Guardian office.

Experiment three: save the MCU

This is where things get serious. For the last decade and a half, the entire global box office has relied on the success of the Marvel Cinematic Universe, at the expense of almost everything else. But just about every new Marvel movie released since Avengers: Endgame has been abysmal, a fact that threatens the entire future of the movie industry. My final prompt was simple: “write me an Avengers movie, featuring Kang as the antagonist and elements of the multiverse in the plot, that is actually good.”

ChatGPT came up with Avengers: Multiversal Menace; a film where Kang attempts to conquer every reality simultaneously. The Avengers split up and travel to different universes to gather allies, but run into problems when they encounter different versions of themselves, unfamiliar technology and “different cultural norms”. And then comes act three:

“The final battle takes place in a multiverse hub, where Kang is attempting to merge all the realities into one to create a new empire. The Avengers and their allies engage in a massive battle against Kang’s forces, while also trying to stop the multiverse from collapsing. The fight is intense and chaotic, with various characters using their unique abilities and gadgets to fight off Kang’s army. In the end, the Avengers are able to defeat Kang and prevent the multiverse from collapsing, but not before they make a shocking discovery about the true nature of the multiverse and its implications for their own reality.”

Again, that sounds like a recipe for muddy, indistinguishable CGI, but why break the habit of a lifetime? Anyway, this is all fine, but I have come to understand that there are only two reasons why anyone goes to see Marvel movies any more: quips and post-credit scenes. I asked ChatGPT to come up with several witty one-liners that the various members of the Avengers might say as they battled Kang, the best of which included:

Thor: “Your conqueror game is weak, Kang. You should stick to playing with your toys.”

Captain America: “You may be a conqueror, but you’ll never conquer our spirit.”

Hulk: “Kang, you wouldn’t like me when I’m angry. Oh wait, you already don’t like me.”

Not great, admittedly, but pretty boilerplate for recent MCU outings. But this led me to my final roll of the dice: an unexpected and breathtaking post-credit sequence. Here, in its entirety, is what it came up with:

“The screen fades in on a dusty, forgotten laboratory in a remote corner of the universe. The camera slowly pans around the room, showing broken equipment and piles of debris. Suddenly, the camera stops on a small, glowing object that’s been hidden under a pile of rubble. As the camera zooms in on the object, the audience sees that it’s a tiny, beeping pager. Suddenly, the pager crackles to life, emitting a faint buzz. The camera lingers on the pager as it buzzes louder and louder, until it finally explodes in a shower of sparks.

“Just as the audience is wondering what this could mean, the camera cuts to a close-up of a pair of hairy, clawed hands. The hands belong to none other than Wolverine, who is revealed to have been standing in the shadows the entire time. With a mischievous grin, Wolverine looks straight into the camera and says: ‘Looks like I’m finally getting that Avengers invite, bub.’

“The scene ends with a classic Marvel-style post-credits tagline: ‘Wolverine joins the fight in Avengers 5: Whatever It Takes.’”

Guys, I think we just saved Marvel. Thank you, ChatGPT.