During the ongoing WGA and SAG-AFTRA strikes, one topic of interest has been AI, or artificial intelligence — really, more accurately, machine learning. Nobody seriously believes that AI is currently in a position to write its own movies with any success, but there have been scenarios floated in which perhaps consumers could use AI-generated scripts and digitally stored copies of actors to essentially choose their own film: "I want a romance starring Ryan Gosling and Emma Stone where she's a bank robber and he's a cop" or something like that. Honestly? Sounds terrible!

Into this landscape comes the movie Choose Love, Netflix's most recent experiment in making "interactive" films. The highest-profile effort up to now was probably the Black Mirror episode "Bandersnatch" in 2018. While it was interesting to see the technology at work (you make a series of choices using your remote, which drives the story forward), it didn't really work narratively. And it doesn't work narratively here, either.

Directed by Stuart McDonald and written by Josann McGibbon, Choose Love is about a woman named Cami (Laura Marano) who works as a recording engineer. She has a boyfriend, Paul, played by Scott Michael Foster. But when she gets a tarot reading because she feels something is missing from her life, she learns that she has three possible suitors. One is Paul. The other two, who quickly turn up, turn out to be her old boyfriend, Jack (Jordi Webber), and a rock star she meets at work, Rex (Avan Jogia). So in one scene (a scene that is not at the end of the movie!), Cami has to decide between Paul, Jack and Rex. And then she ends up with one of them. That's the story.

Now, you might ask yourself why a woman with a lovely boyfriend would suddenly leave him for either an old boyfriend she's been apart from for years or a rock star she met at work with whom she's spent a few hours. I asked myself that too! That's why Paul was my choice, so I picked Paul, so Cami picked Paul, and she ended up with Paul, and ... there you go. I did not watch every minute of the other two ways for the story to go (it doesn't really lend itself to any particular linear or completist viewing in any handy way), but I explored the other possible storylines enough to learn that there's no particular cleverness — it's not as if you pick one but end up with another one, or no matter who you pick, you end up with the same guy, or something like that. You pick the ending you want, and it gives it to you. I didn't count, but I would say I probably made ... a total of 15 or so choices over the course of the movie? Something like that? Some of them matter a little, some matter almost not at all.

It's profoundly unsatisfying, and I think it conceptually fails to understand how much of fiction is about letting a creative person make creative choices and experiencing them as a viewer/reader. But I also think it fails to understand that interactive storytelling already has some very successful models, and they don't look like this.

They look like the wonderful and haunting video game Kentucky Route Zero, a seemingly simple point-and-click game over the course of which you make many, many choices about where to go next, what to say, what to do. What you do not do is get an endings list from which you get to choose. There is one ending. You will get there. The same is true of the beloved game The Last of Us, which not only has a single ending, but has a single controversial ending. That's not to say there aren't beloved games with multiple endings; there absolutely are. But successful interactive storytelling has models to learn from, and they generally involve iterative choices that take you down branching paths, not "which guy do you want to end up with" questions that just kind of ... serve you whatever outcome you ordered.

That's why this might have worked if your choices had not gone as you anticipated — picking Paul in that moment led you to end up single, something like that. But this? Picking an end point and watching yourself get there because ... you asked to get there? If people want to do that, they can just buy a set of finger puppets and play the thing out at home. If you want people to write their own endings, you're getting into the realm of fanfiction, and fanfiction requires people to have a lot more creative options, a lot more ways to go. In other words, you can write the ending of your story or I can, but for you to write a menu and me to order from it feels like an uncanny valley of creation that satisfies nobody. I mean, I didn't think Ross and Rachel should have ended up together on Friends, but do I want to watch an AI-generated ending where she doesn't get on the plane? No! What would be the point?

This is why I cannot get excited about the idea of a bespoke, AI-generated movie that I order up like a pizza. The enjoyment of fiction lives in the meeting of your mind as a reader or viewer with the specifics of other people's brilliant, weird, flawed, unexpected minds. Even in a genre with rules like murder mystery (the crime will be solved) or romance (there will be a happy ending), there is always discovery. Without discovery — without the possibility of frustration, in fact — there are no stakes in the act of reading or watching.

There are a lot of ways that machine learning could harm actors and writers; that's why it's an issue in the strikes. But any idea that boundless customization is the way of the future seems to me to misunderstand the relationship between creator and audience and the negotiation that goes on between the two. When I watch your movie, I want to like it. You hope I like it. I know I might not. You know I might not. But you have your role to play, and I have mine, and we both hope for the best. It's terrifying, but the joy of fiction lives in that uncertainty, and it's not a problem for technology to solve.

This piece also appeared in NPR's Pop Culture Happy Hour newsletter. Sign up for the newsletter so you don't miss the next one, plus get weekly recommendations about what's making us happy.

Listen to Pop Culture Happy Hour on Apple Podcasts and Spotify.

Copyright 2023 NPR. To see more, visit https://www.npr.org.

300x250 Ad

300x250 Ad

Support quality journalism, like the story above, with your gift right now.

Donate