AI enters filmmaking: hundreds of film crew jobs at risk

AI now shoots films and puts hundreds of film crew jobs at risk

On October 30, a full-length film made entirely with artificial intelligence is scheduled to premiere in some theaters. The work of a creator known as Gossip Goblin, who already publishes AI-made science fiction videos, the news, reported by IndieWire, is the tip of the iceberg in a debate sweeping the internet: if an author with a computer can manufacture a blockbuster, what happens to the dozens or hundreds of people who make a conventional shoot possible?

The question is not idle. In production alone, there is an executive producer, producer, co-producer, production manager, department heads and assistants, coordinator, secretarial staff, accountant, and auxiliaries. Then come directing, screenwriting, cast, casting, cinematography, electricians, grips, sound, art direction, costume, makeup, and hair. The organizational chart exceeds fifty trades. AI does not promise to wipe them all out at once. It does promise to make many unnecessary.

How many people work on a professional shoot

The structure of a shoot is not capricious: each position exists because there is a task that cannot be solved with a camera and good will. Some recall being surprised to see a film shot in their town, with a deployment they describe as an entire army mobilized to film a scene of a few seconds. That army is exactly what AI puts in its crosshairs.

---

The industry does not deny the advance but debates its scope. Some argue that AI will not wipe out the arts and that quality cinema will continue to exist, albeit with much smaller budgets. Others reply that classic filmmaking depends on multi-million-dollar budgets that AI makes unnecessary, and that is the real earthquake: not replacing the director, but eliminating the barrier to entry.

From a 1968 film to the Will Smith spaghetti video

It is worth looking back to measure the leap. That 1968 film cost 12 million dollars and more than half went to special effects. It is said that the techniques to achieve certain believability were so artisanal that today they would provoke disbelief. That cinema required huge teams for a single shot. The viral video of a deformed Will Smith eating spaghetti seemed like a joke. Today, models generate action scenes that, for some participants, are not very different from much of current superhero cinema.

The comparison is not gratuitous. Defenders of traditional cinema say there is still a difference, and that watching an entire AI-made film gives you a headache. Enthusiasts reply that it is only a matter of time.

Why detecting AI content is not enough

The human eye improves, but not at the machine's pace. The average accuracy for detecting AI-generated content is around 50 to 60%, which is equivalent to flipping a coin. And that is when the model still has coherence failures. The practical consequence is uncomfortable: as one participant summarizes, if you cannot distinguish, you cannot choose either.

---

Some point to a specific case: turning the phone screen while watching a series, they discovered that several secondary characters moved strangely when walking, except the protagonist and one or two others. The suspicion remained there, unproven. But it is precisely the kind of detail the industry will have to manage.

The model that trains on what it has made itself

Here the technical problem that almost no one discusses aloud emerges. Models were trained on material made by humans. Now, according to one participant, much of the new material circulating is also generated by AI. Whoever feeds new videos to the system finds that a growing fraction did not come from a camera. There would be more and more generated content and less filmed content. What happens when the model feeds on its own harvest is the question no one fully answers yet.

There will be actors, but scanned forever

The list of trades does not disappear entirely, some argue. A scout will find a pretty face and, instead of setting up an acting course, will offer a million for their image for all eternity: 3D scanner and shoot with their face until the end of time. No more changing actors because they no longer look like a teenager. You can make a thousand films with the same princess even if each one takes ten years to release.

A settling of scores with the industry

Part of the enthusiasm is not technological, it is settling scores. One current holds that the audiovisual sector concentrates a very specific ideological profile and that AI is going to run over it. Another celebrates the end of the characters from the culture world who, in their view, never had to study. These are opinions, not data, and should be taken as such.

What is real is the desire to bring to the screen stories Hollywood would never produce. There is talk of Tartessos, Numancia, Covadonga, the Roman era, Lepanto, Las Navas de Tolosa, the tercios, or the siege of Malta. If the barrier to entry falls, they say, any good story can be shot.

The problem, admit the most lucid, will no longer be producing. It will be separating the wheat from the chaff. And there, if something is easy to do, it loses value. The discussion gets stuck exactly at that point.