Will Taxi Driver 2 be written by AI?
What: 5 Loglines Showdown
When: Friday, January 30
Deadline: Thursday, January 31, 11pm Pacific Time
Submit: 5 loglines, each with a title and a genre
Where: carsonreeves3@gmail.com
So, the other day, I heard that Paul Schraeder was in the news for saying outlandish things. For those of you who don’t know, Paul Schrader is a famous Hollywood screenwriter best known for scripting Taxi Driver. He also wrote Raging Bull.
Just to provide some context, Schrader’s last five movies were Master Gardener, There Are No Saints, The Card Counter, Dark, And First Reformed.
The reason, it turns out, Schraeder was in the news is because he said he’s been spending a lot of time on Chat GPT and that the AI program is better at coming up with movie ideas than he is.
Now, the last time I checked this, which was about six months ago, that statement would be demonstrably false. AI could come up with these ideas that SOUNDED like movies. But they were always missing an ingredient or two that made them sound like an actual movie.
Maybe that’s changed. But that was my experience six months ago.
Another thing to keep in mind here is that Paul Schraeder isn’t exactly an ideas guy. Taxi Driver is literally a movie about a guy driving around in a taxi. Its strength is in its execution, not in its concept.
And here’s the logline for Master Gardener: “A meticulous horticulturist is devoted to tending the grounds of a beautiful estate and pandering to his employer, the wealthy dowager.” Not going to win any Pitch Fests, that’s for sure.
So, from the jump, my initial thought was, “Well, Paul, lots of people could come up with better movie ideas than you.” He’s not very good at it.
But, in the leadup to next week’s Logline Showdown, I’m curious where we stand on AI movie ideas. Is it getting to a point where AI could actually come up with an idea that’s worthy of spending 30 million dollars on?
So, what I thought I’d do is challenge AI to come up with five movie ideas in the five most marketable genres. Horror, Thriller, Action, Adventure, and Sci-Fi. I’m going to tell it exactly what I’m telling you here. In fact, I’m going to paste this paragraph into the prompt and tell it I want it to come up with its five best loglines for those genres. And then I want to dissect those ideas and see if they’re any good.
Are you ready?
Okay. Here goes nothing!
Horror:
Echoes in the Hollow – When a grieving mother moves into an isolated mountain village, she discovers the town’s eerie folklore about soul-stealing shadows might be true, as she begins to hear whispers from her dead child calling her into the woods.
Thriller:
The Algorithm – A brilliant tech investigator uncovers a government-run AI capable of predicting crimes before they happen, but as the AI starts framing innocent people, she must outsmart the machine before she becomes its next target.
Action:
Code Red: Omega – When a rogue paramilitary group hijacks a U.S. missile submarine, an ex-Navy SEAL turned disgraced shipwright must infiltrate the vessel to prevent a global catastrophe, relying on his engineering wits and old combat skills.
Adventure:
The Titan’s Map – After discovering a hidden map etched into an ancient statue, a group of rival treasure hunters embarks on a perilous journey across the globe, chasing a mythical lost city rumored to contain an artifact that can rewrite history.
Sci-Fi:
Quantum Divide – In a future where teleportation has revolutionized travel, a physicist discovers a dark secret: each teleportation creates a copy while killing the original. When the technology is weaponized, she must expose the truth before humanity destroys itself.
First impressions?
Before I share those, let me emphasize that the ultimate goal of a logline is to make the reader go, “Holy shit, I want to see that movie now.” It’s rare to achieve that, whether you’re human or artificial intelligence, but that’s the goal.
So, that’s the initial criteria I’m judging these loglines by. And none of them pass the test. Nothing here makes me go, “I have to see this!”
With that said, none of them are bad. I didn’t read any of these loglines and roll my eyes, which, to be fair, is something I do often when I go through logline submissions. Just the fact that there are no spelling or grammar errors puts them above 60% of the entries in my competitions.
But, ironically, AI’s loglines have the same issue that a lot of human loglines have, which is that they sort of feel like a movie you’ve already seen, yet they don’t offer a fresh enough perspective to make this new version movie-worthy.
Let’s take a look at them one by one.
Title: Echoes in the Hollow
Genre: Horror
Logline: When a grieving mother moves into an isolated mountain village, she discovers the town’s eerie folklore about soul-stealing shadows might be true, as she begins to hear whispers from her dead child calling her into the woods.
The “special attractor” here (otherwise known as the variable in your story that’s supposed to make your idea sound exciting) is “soul-stealing shadows.” I’m just not interested in shadows. I don’t find them very compelling. In fact, they seem kind of boring. Shadows? So, your special attractor is weak from the get-go.
Then, you have the dead kid calling from the woods trope. I mean, come on. I must’ve read a thousand scripts where you can hear kids or babies, laughing or crying from the woods or deep in the house. It’s lazy and it’s far from a hook.
Again, there’s nothing wrong with this idea. But nothing about it stands out. Which means it’s going to get passed over.
Title: The Algorithm
Genre: Thriller
Logline: A brilliant tech investigator uncovers a government-run AI capable of predicting crimes before they happen, but as the AI starts framing innocent people, she must outsmart the machine before she becomes its next target.
I’ve noticed that AI does this thing where it takes movies that already exist and makes a minor change to them. This is just Minority Report with AI instead of those milk-laden prognosticator people.
But, to be fair, this is what humans do, too. They keep trying to spin their favorite movies in new directions. The success of a movie idea often lies in finding the perfect balance—pushing the concept in a fresh, new direction while still preserving the core elements that made the original idea compelling.
The Algorithm is too close to the original Minority Report to feel fresh.
Title: Code Red: Omega
Genre: Action
Logline: When a rogue paramilitary group hijacks a U.S. missile submarine, an ex-Navy SEAL turned disgraced shipwright must infiltrate the vessel to prevent a global catastrophe, relying on his engineering wits and old combat skills.
This is an interesting one because straight action concepts aren’t usually flashy. With that said, Hollywood still places value on originality and a flashy concept in a genre they love to make could be the difference between your movie getting picked over another one.
I know that Hollywood has been looking for a great sub script for a while now. Therefore, if you can come up with a good submarine concept, you have a chance at filling that requirement.
Ironically, the problem with this logline is a logistical one. Which I didn’t expect to encounter with AI, since AI is built on 1s and 0s. But how do you infiltrate a submarine that someone has already hijacked? Do you swim down there and, when they’re not looking, climb in through the missile hole?
I suppose he could’ve already been on the sub. And, if that’s the case, it’s not a bad idea. I would’ve liked a little more irony in his job (having him be a ship builder feels barely adjacent to the task at hand).
And then the ‘save the world’ thing feels generic. But I guess a lot of big action movies subscribe to those stakes. — On the plus side, I like the title.
Title: The Titan’s Map
Genre: Adventure
Logline: After discovering a hidden map etched into an ancient statue, a group of rival treasure hunters embarks on a perilous journey across the globe, chasing a mythical lost city rumored to contain an artifact that can rewrite history.
More than any other example, this idea feels stitched together from other movies. And this is the big complaint against AI, that it doesn’t know how to create on its own. Its creation is always built on top its database of previous creations.
And while the argument is that humans do the same, I would argue they go about the process differently. AI literally cherry picks variables from these past successful movies and stitches them together in a mechanical way.
Humans tend to think more emotionally when they build on top of previous ideas, utilizing what they feel, specifically in relation to how humans would be involved in these stories.
Most of the time, protagonists are stand-ins for the writers writing them. Writers use the characters they’re writing to work through their own problems they’re dealing with in life. And that’s what makes the story feel relatable and human.
AI has not figured out how to do that. Nor do I think it’ll ever figure that out. Because how do you work through a human experience if you’ve never been a human?
On top of that, The Titan’s Map feels like an old discarded Indiana Jones sequel pitch. It’s just not very original.
Title: Quantum Divide
Genre: Sci-Fi
Logline: In a future where teleportation has revolutionized travel, a physicist discovers a dark secret: each teleportation creates a copy while killing the original. When the technology is weaponized, she must expose the truth before humanity destroys itself.
Same problems. AI is aware of Christopher Nolan’s The Prestige. It’s using the basic premise behind that as its foundation. And the rest of the idea doesn’t even make sense. It’s going to be weaponized??? What do you mean? Are you going to trick people into teleporting so you can secretly kill them?
I’m just working this out in my head here. Teleportation R&D, building of the devices, powering these devices – probably equates to a couple of million dollars per teleport? How much does a bullet cost again? About 10 cents?
Can someone do the math for me? Which one of those options is more cost-efficient if you want to kill someone?
Look, maybe the implementation of the weaponization makes more sense in the actual script. But it’s your job as the writer to convey that in the logline. Or else you’re going to get snarky responses like this one.
For fun, I copied and pasted my critiques of all five loglines back into ChatGPT and asked it to rewrite the loglines incorporating the criticism. I’m not going to include all of those rewrites since I don’t want this post to be 5000 words. But I’ll include one to give you an idea of if it’s able to improve.
Title: The Titan’s Map
Genre: Adventure
New Logline: When an estranged father and daughter uncover a map etched into an ancient artifact, they must navigate deadly rivals and their fractured relationship as they search for a mythical city rumored to grant its finder unlimited power.
As you can see, AI tries to create an approximation of a “human issue,” by adding a father daughter team with problems. But it’s conveniently just that – the symbol of a human problem. There is no sense at all that these two are real people. That’s what AI continues to miss the mark on.
So, what’s my conclusion in all of this?
AI has gotten A LITTLE BETTER at creating movie ideas. But because it doesn’t have any power to create brand new ideas out of nothing and because it lacks any understanding of human behavior, its output continues to result in half-baked ideas without any emotional depth. The more you scrutinize them, the less they hold up.
