Dreaming of Movies that Might Actually Get Made

I graduated from the Netherlands Film and Television Academy, as it was called at the time. That probably sounds more important than it really is because I doubt that they ever not graduated any of the students. Dutch director Paul Verhoeven, who did not attend that school, once pointed out to me, half-jokingly, that it shouldn’t be called Academy because it’s a trade school and not part of academia. Verhoeven ( Basic Instinct, RoboCop ) himself has degrees in mathematics and physics.

Do the developments in AI mean that I could write a script and actually make it into a movie? Produce whatever I want however I want without the usual gatekeepers deciding what is permissible. Is AI going to suck me into another identity wormhole? Let me not get my hopes / fears up prematurely.

I did work in film and TV in the Netherlands for a while so I do know the industry, at the Dutch scale at least, from the inside. AI is reshaping the film industry profoundly. AI has been able to generate images for quite a few years. Put a whole bunch of images together in a sequence that makes sense together and voila: we have video. The quality of AI generated video is so good now that most people can’t tell if it’s shot with a camera or if it originates from a prompt brought to life in a datacenter. That includes dialogue between several characters.

Producing a movie the traditional way costs on average something like 50 million dollars a pop. Making one using AI of similar production quality would cost about half a million. So that’s 1%. That half a million budget would not include any A-list actors, unless they would license their likeness to the producers, which would add to the budget.

The half a million budget would mostly consist of wages for a script writer, an editor, a colorist, an audio designer and a producer. A lone multi-talented filmmaker would just invest his time and the cost of generating footage. Roughly calculated that would be $10 for every generated minute, times 1000, so that’s $10,000. This assumes a 1:8 ratio for footage to a final length of 120 minutes, a common movie length.

At the time of this writing AI can generate footage really well, but as far as I know the same doesn’t hold true for scriptwriting. I have been following author and pundit David Shapiro with growing interest. He regularly writes about his efforts to write a novel using AI. He reports that they struggle with theory of mind. A lot of the dialogue comes out as if it’s all said by the same person. He uses the latest frontier models, so it begs the question if this is something that AI will never be able to do or that in time it will acquire that skill as well. Based on how things have been going, my guess is the latter.

So what will be left for us after that? Taste? Anything else? The weird dynamic of AI is that while with top quality generative video it may give millions of people the opportunity to do something really creative, at the same time it’s moving ahead to absorb skills that are closer to the center of what we think makes us human. Where will that end? What if in a few years the best movies will be written, directed and edited by AI? Or will there be a point where it just can’t move any further? If so that could be an important demarcation point telling us who and what we really are. That is, if we choose to define ourselves by our skills.

Almost anyone will be able to make feature length movies, but that of course does not mean everyone will also automatically have good taste and artistic judgment. A common mistake I see is overdoing it for things that used to be very difficult and expensive to produce before AI. An example of that in my opinion is Primordial. But hey, if they find an audience who loves a gluttonous helping of special effects, who am I to have a problem with that. I expect in a few years we will look back on it as a clearly dated early AI style.

At the moment there are no feature-length AI-generated movies scheduled for theatrical or Netflix release. With AI-generated in this context I am referring to movies of which all footage is AI generated, including the actors. There are plenty of movies of which part of the footage is AI generated. Netflix reported that 300 of their productions included AI-generated footage. When will we see AI actors in theaters and on Netflix? It could be next year or 2028. The technology exists today but there is understandable industry resistance.

If movies will only cost 1/100 or even less of the usual price tag, will there also be 100 times as many movies? What will this mean for the distribution?

Start thinking about movies somewhat similar to the way we think about books. Anyone can write a book, soon almost anyone can make a visually studio-quality movie. This will result in a greater emphasis on marketing. Like books today, many movies will be made that only a few people may ever see. In the transition period the established industry will have inertia and the solo cineastes can take advantage of that.

To distribute this many movies Amazon is positioned better than Netflix, both with technology and mindset. Amazon just needs to slightly adjust what they already have for books and they own large data centers to store the hundreds of thousands of movies. Netflix needs to redesign how they present their catalog to the user and needs to re-invent itself. They did that before when they went from sending DVDs through the mail to streaming but they are no longer the small nimble startup.

All in all, I am looking forward to where this is going. I really don’t like most movies that are released these days. The two second shots, the shaky camera, the disjointed storytelling. Will AI be the new 16mm Beaulieu movie camera? We may see a resurgence of auteur cinema, a cinematic movement that started the in 1960s and included the French New Wave. I will find those movies just like I find books today. (Okay, sure French New Wave movies also have their two second shots, shaky camera and disjointed storytelling. I just like them better in Breathless than in Oppenheimer.)

And for me, I can dream about movie ideas again. This time with more than a snowball’s chance in hell that my script will ever materialize into a movie.