I hate AI. Let’s say that right up front. I hate chatbots helping me resolve my issue, machines that “learn” my preferences, automated call centers that scold me with an emotionless “I’m sorry you’re having trouble.” I’m sad that you won’t see this article unless I appease Google’s AI crawlers by frequently slipping in the search term “AI-generated screenplay.”
I once talked to a therapist about how uncontrollably angry I get at these automated methods of herding people’s behavior and she told me, “No — you’re right to be furious.”
So I’ve followed the fears about AI taking over screenwriting jobs with great interest. It seems like something that could happen, right? Algorithms curate what we watch on streaming services, what songs appear on our playlists, and what ads pop up on websites (including this one). YouTube has learned to guide people down hellish rabbit holes. How hard would it be for something like ChatGPT to write a script? (Excuse, I mean “an AI-generated screenplay.”)
With Hollywood writers on strike, some executives are reportedly already toying with the idea of putting A.I. to work. These are presumably the same types of executives who say things like “make a hit movie” or “create a viral video” with no understanding of how much work or luck is involved.
So not long ago, before the strike, I became one of the countless people who have tested ChatGPT with a list of amusing-to-me commands. For my first trick:
I Asked ChatGPT to Write the Screenplay of an Oscar-Winning Movie
I gave it free rein, and no parameters or prompts. Nothing here is from me except a request to write an Oscar-calibre screenplay. Not a single thing.
So… here’s what it wrote.
EXT. A RUGGED MOUNTAINOUS LANDSCAPE – DAY
The camera pans across a breathtakingly beautiful terrain of towering peaks and cascading waterfalls. We hear the faint sound of Tibetan music in the background.
INT. A SIMPLE MONASTERY – DAY
We see a group of monks sitting in silent meditation, their faces serene and peaceful. One of the monks, a young man named TASHI, stands up and walks towards a shelf filled with ancient scrolls. He selects one and begins to read.
EXT. A REMOTE VILLAGE – DAY
Tashi walks into a small, impoverished village where the inhabitants live in extreme poverty. He approaches an elderly woman and hands her a bowl of rice.
TASHI (in Tibetan, subtitled) May you be blessed with abundance and prosperity.
The woman nods gratefully and Tashi walks away, his eyes filled with compassion.
EXT. THE MOUNTAINS – DAY
Tashi is seen walking up a steep mountain path, his feet sinking into the snow. He is carrying a large bundle of supplies on his back.
INT. A SMALL CAVE – DAY
Tashi enters a dimly lit cave and approaches an old man, who is lying on a bed of straw. The man is emaciated and frail, his breathing labored.
TASHI (in Tibetan, subtitled) Master, I have brought you some supplies.
The old man nods weakly and Tashi begins to feed him, spoonful by spoonful.
EXT. THE MOUNTAINS – DAY
Tashi is seen sitting atop a mountain peak, gazing out at the stunning vista before him. The camera pans out and we see him as a small speck against the vastness of the landscape.
Me again. Hi! That’s all it wrote. Tashi doing a bunch of Save the Cat things. If you Google around, and you’ll see that ChatGPT seems to have potentially borrowed/stolen from lots of real things. But the main problem is that the above script, while properly formatted and written grammatically, is boring.
Some Preliminary Thoughts on That, Ahem, AI-Generated Screenplay
But I asked ChatGPT to continue. If you’re already bored, feel free to scroll down past the bold text for some takeaways, conclusions. But if you have the patience of Tashi, read on:
EXT. THE MOUNTAINS – DAY
The wind howls as Tashi stands on the mountain peak, his eyes closed in deep concentration. He takes a deep breath, opens his eyes, and begins to chant.
TASHI (in Tibetan, subtitled) Om mani padme hum.
As Tashi chants, the camera pans out to reveal a group of monks standing on nearby peaks, also chanting in unison. The sound of their voices fills the air.
INT. THE SMALL CAVE – DAY
The old man is lying on his bed, his breathing shallow and labored. Suddenly, his eyes flutter open and he looks directly at Tashi.
OLD MAN (in Tibetan, subtitled) You have fulfilled your duty, my son. It is time for you to embark on your own journey.
Tashi nods, tears streaming down his face.
TASHI (in Tibetan, subtitled) Thank you, Master. I will continue to serve others as you have taught me.
EXT. THE REMOTE VILLAGE – DAY
Tashi walks through the village, his eyes scanning the faces of the people he passes. He stops in front of a small hut and knocks on the door.
A YOUNG GIRL opens the door and looks up at Tashi with wide eyes.
TASHI (in Tibetan, subtitled) Is your mother home?
The young girl nods and runs inside. Tashi waits outside, his eyes scanning the village.
INT. THE SMALL HUT – DAY
Tashi sits with the young girl’s MOTHER, who is sick and weak. He feeds her a bowl of rice and begins to speak to her in a gentle voice.
TASHI (in Tibetan, subtitled) You are not alone, my sister. We are all connected, and we will all find peace together.
The mother nods, a small smile on her face. Tashi stands up and bows to her.
TASHI (CONT’D) (in Tibetan, subtitled) May you find peace, my sister.
EXT. THE REMOTE VILLAGE – DAY
Tashi walks away from the hut, his eyes scanning the village. He pauses for a moment and looks up at the sky, a sense of peace and purpose washing over him.
That’s it! Still no conflict. No drama. I asked ChatGPT to keep going. Here we go:
As he walks, he pauses to take a sip of water from a canteen. His eyes scan the mountainside, taking in the beauty of the landscape around him.
EXT. A SMALL MOUNTAIN VILLAGE – DAY
Tashi arrives at a small mountain village, where he is greeted by a group of children playing in the dirt. They stare up at him, their eyes filled with curiosity.
Tashi smiles at them and reaches into his backpack, pulling out a small flute. He begins to play a gentle melody, and the children gather around him, listening intently.
INT. A SIMPLE HUT – DAY
Tashi sits with a group of villagers, listening to their stories of struggle and hardship. They tell him about the lack of food, clean water, and medical supplies, and how difficult it is to survive in the harsh mountain environment.
TASHI (in Tibetan, subtitled) I understand your suffering. But we must remember that we are all connected, and that by helping each other, we can create a better world.
The villagers nod, their eyes fixed on Tashi with hope and respect.
Me again. You’re bored, right? You’re probably not even here anymore. I asked ChatGPT to keep going, and repetition started to creep in — not just the redundancy of Tashi being nice, but also redundant phrases. (Tashi often has his “eyes fixed on a distant peak.”) So…
I Asked ChatGPT to Come Up With a Big Twist
After a lot more of Tashi walking around, staring and being nice, this is what ChatGPT came up with:
Suddenly, a group of armed men emerge from the surrounding mountains, their faces covered with masks.
TASHI (in Tibetan, subtitled) Who are you? Why have you come here?
LEADER OF THE ARMED MEN (in Tibetan, subtitled) We are the Red Hand, and we have come to take what we want. Hand over your supplies and we’ll let you live.
Tashi stands tall, his eyes fixed on the leader of the armed men.
TASHI (in Tibetan, subtitled) I will not let you take what we have worked so hard to build. We are all connected, and by helping each other, we can create a better world.
The armed men laugh, but Tashi remains steadfast. Suddenly, a group of monks appear from the surrounding mountains, their faces serene but determined.
MONK (in Tibetan, subtitled) You will not take what is not yours. We will defend this village with our lives.
A fierce battle ensues, with Tashi and the monks fighting against the armed men. The villagers join in, picking up whatever weapons they can find.
After a long and grueling fight, the Red Hand is defeated. Tashi and the monks stand tall, their eyes fixed on the horizon.
TASHI (in Tibetan, subtitled) We will continue to help each other, no matter what challenges we may face. We
It paused. I jumped in with some encouragement: “keep going this is great.” ChatGPT continued with the, yes, “AI-generated screenplay”:
Suddenly, a figure appears from behind the armed men. It’s Tashi’s long-lost brother, Sonam, whom he thought had perished in a mountaineering accident years ago.
OK… so. Not so great. First, “whom he thought had perished in a mountaineering accident years ago” is a classic tell-don’t-show mistake, since the audience will be watching, not reading this exposition in the script. But look, it tried. I also like The Red Hand as the name of the bad guys.
But, I was generally bored, so I asked ChatGPT to liven things up with a visit from Dave Chappelle. ChatGPT invented a character named “Darnell.”
Suddenly, a man walks into the village. It’s Darnell, a wandering comedian, played by Dave Chappelle.
DARNELL (to Tashi) Hey man, what happened here?
TASHI (sadly) We were hit by an avalanche. Many homes were destroyed, and we lost many lives.
DARNELL (nodding) That’s rough, man. I’m sorry to hear that.
Me cutting in again: I can kind of picture the Dave Chappelle character saying this.
TASHI (curious) What brings you to our village, Darnell?
DARNELL (smirking) Well, I’m a traveling comedian, and I heard about this place. Thought I’d stop by and see if you guys could use a good laugh.
Darnell turns out not to be a great guy, sadly. I also asked for “a character who could be played by Cheryl Hines” and she was cast as a journalist who helps expose Darnell as a thief, and her editor tells her that her big article, which will come out in a few weeks, is going to be submitted for a Pulitzer. Which seems presumptuous, since it hasn’t come out yet, but who am I to judge?
I asked ChatGPT to keep going, informing it that I needed at least 120 pages, asking for more, and this happened:
CHERYL (confused) That’s strange. My editor just messaged me again. He says there’s been a mistake.
TASHI (curious) What kind of mistake?
CHERYL (reading the message) He says that they made a mistake with the article. They’re not going to be submitting it for a Pulitzer after all.
SONAM (surprised) What? Why not?
Cheryl looks up at them, a serious expression on her face.
CHERYL (grimly) Because Darnell bought the Pulitzer committee.
Tashi and Sonam look at each other, stunned.
TASHI (disbelieving) What? How is that even possible?
ChatGPT couldn’t sustain 120 pages, I’m afraid, and no one is still reading this anyway. But it did come up with this hopeful conclusion:
ChatGPT wasn’t able to keep things going for 120 pages, sadly, but did wrap things up on a hopeful note, with a voiceover from Cheryl:
CHERYL (V.O.) But in the meantime, we can celebrate the victories. We can appreciate the moments of joy, and we can hold on to the hope that someday, justice will be served.
The camera lingers on Tashi and Sonam, standing together and watching the happy scene.
CHERYL (V.O.) Because that’s what it’s all about. Not just the big victories, but the small ones, too. The moments of connection, the acts of kindness, the little things that remind us that we’re all in this together.
The camera fades to black, and the credits roll.
So, some takeaways.
First, if you’re an executive of the “make it go viral” variety, please understand this core point, if nothing else:
What ChatGPT has written above is not very good.
I’ve asked ChatGPT to do a few other exercises, including writing scenes for a script started by a real, incompetent human (me). Some patterns emerge: redundancy, predictability, twists we’ve seen many times before, things that don’t always feel logical.
Of course, this happens with human scripts as well. Most stories follow a similar format, and many follow similar formulas, and many have the same old reveals, twists and surprises. What makes great scripts great are things that feel authentic, surprising, and human. Great films pull us in so completely that we forget we’re watching a work of fiction, carefully crafted by deeply committed experts in their field.
Take CODA, for example.
Why CODA Works
On one hand, the 2022 Best Picture winner, CODA, might sound formulaic: it’s a movie about a child of dead adults who wants to escape her small town through music. Yet when you sit down and watch it, the movie feels alive, and gets you deeply invested in its characters, because writer-director Siân Heder went to such lengths to spend time with the kinds of people she portrays, and to understand the desires and details that make them human. (On top of that, you have outstanding acting, music, and cinematography — things the script can’t fully control.)
ChatGPT may work best not as a generator of new ideas, but as a barometer of what’s already been done to death. It can’t replace screenwriters — it tells you why movies need screenwriters.
If you’re a writer who asks ChatGPT to finish a scene for you, and it comes up with the same conclusion you had in mind, the odds are good that your approach isn’t surprising, but is rather so ingrained in storytelling formulas that a mindless machine can see it coming in any, you guessed it, AI-generated screenplay.
But at least in the case of Tashi, it’s also interesting to see what ChatGPT thinks people want to see: calm, patience, people helping other people. There’s something nice about that. It’s possible that it can see the best of us.
Or that it’s simply designed to make us see the most reassuring versions of ourselves, like so many algorithmic bubbles.
Yes, there are things ChatGPT can do, like research. But it just can’t be as unpredictable or convincingly surprising or suspend your disbelief like a script by a human.
In the end, that’s what it’s all about. Not just the big victories, but the small ones, too. The moments of connection, the acts of kindness, the little things that remind us that we’re all in this together.
Main image: A Shutterstock of a robot, perhaps writing a — we’re so sorry — AI-generated screenplay.