What is Gore Verbinski's proposed AI rating system for films?

Gore Verbinski proposed a gradient rating system at the Taormina Film Festival that grades filmmakers based on how much AI was used in their production. An F would be given for AI-written scripts, while higher grades could be earned when AI is used transparently for specific creative purposes, such as generating imagery that serves an emotional metaphor and could not otherwise be afforded. The system emphasizes transparency about what the tool did and did not do.

How is AI replacing apprenticeships in the film industry?

AI is primarily taking over entry-level and menial tasks that were historically performed by interns, apprentices, and assistants — positions that served as de facto education in filmmaking craft. Grips, production assistants, and assistant editors traditionally learned by physical proximity to the work: watching lighting setups, observing camera blocking, syncing dailies. When AI automates these tasks, the pipeline that produces skilled filmmakers with deep craft vocabulary shrinks, because the jobs that taught those skills no longer exist.

Can a rating system effectively measure AI use in filmmaking?

Rating systems and certification marks face a fundamental limitation: they measure how a film was made, not whether it is good. A grade certifies process, not quality — a lazy hand-written screenplay and a masterpiece both receive the same high mark. Like the Human Made Mark, Academy authorship standards, and copyright frameworks, Verbinski's letter grades would evaluate the kitchen without tasting the food. The deeper challenge he identifies is whether future filmmakers will develop enough craft vocabulary to receive a meaningful grade at all, given disappearing apprenticeship pathways.

You get an F -- CinePrompt Field Notes

Gore Verbinski told the Taormina Film Festival this week that the film industry needs a rating system for AI. "If you use AI to write a script, you get an F." He said it casually, the way someone describes an obvious solution to a problem nobody has solved. The audience laughed. The sentence traveled.

Verbinski is not a purist. He directed three Pirates of the Caribbean films, a franchise that employed more CGI artists than sailors. He noted that AI-adjacent tools have existed in filmmaking for twenty years, embedded in grading software, sharpening algorithms, visual effects pipelines. The checkbox on the submission form that asks whether AI was used has already become, in his words, "very complicated." The technology is in the water supply. The question is how much.

His rating system proposes a gradient. Not a binary but a spectrum. An F for AI-written scripts. Something higher for a filmmaker who uses AI to generate a passage they could not afford to shoot, one that serves what he called "a central aspect" of the emotional metaphor. That usage is "ok." The condition: absolute transparency about what the tool did and did not do.

The gradient acknowledges degrees. A filmmaker who hand-wrote every word and then used generation for ten seconds of surreal imagery is not doing the same thing as a filmmaker who pasted a logline into a chatbot and received a screenplay. The binary cannot distinguish between them. The gradient tries.

But letter grades measure input, not output. An F for an AI-written script says nothing about whether the script is good. An A for a fully hand-crafted production says nothing about whether the film is worth watching. A lazy screenplay typed over three weekends gets the same A as Chinatown. The grade certifies process, not quality. This is the same epistemological problem every institutional response to AI filmmaking has encountered. The Human Made Mark certifies presence. The Academy certifies authorship. Copyright certifies decision-making. The EU certifies oversight. Each measures a dimension of how the work was made. None measures whether the work was any good. The rating system would join a shelf of instruments that evaluate the kitchen without tasting the food.

Verbinski knows this. He is not naive about the limitations. What makes the Taormina interview worth reading is not the grades. It is the apprenticeship observation, delivered almost as an aside, that contains the sharper diagnosis.

"The loss of apprenticeship is a major concern," he said. AI has primarily taken over menial tasks formerly attributed to interns, apprentices, and assistants. He sees it happening in law firms, in creative agencies, everywhere. Entry-level jobs are disappearing. And entry-level jobs were never just labor. They were education.

The grip carried C-stands for two years before anyone let her touch a light. In those two years she watched a hundred setups. She learned where the DP placed the key by proximity, not instruction. She learned the difference between motivated light and unmotivated light by standing in both and noticing how the room felt different. Nobody taught her. She was there, and the work taught itself through her hands.

The production assistant who became a director watched forty camera setups a day while running errands. Nobody invited him to observe blocking. He observed it because the blocking was happening in the room he was already in, carrying a clipboard and a radio. The knowledge was ambient. It transferred through presence, not curriculum.

The assistant editor who became an editor synced dailies for eight months. Tedious, mechanical, low-paid work. But eight months of watching every take from every angle of every scene in a production is eight months of absorbing editorial rhythm at a level no film school provides. The menial work was the masterclass.

AI does not replace the master. It replaces the apprentice's desk. The desk where the next master was going to sit and learn by doing the boring part until the boring part became instinct.

This is the supply-side version of the vocabulary problem. The gap between what a filmmaker knows and what a model comprehends is only meaningful if filmmakers continue to develop the knowledge that makes the gap navigable. Vocabulary does not emerge from a prompt. It emerges from years of physical proximity to the work. Watching someone light a set. Watching someone build a frame. Watching someone choose a take. Doing it wrong a hundred times until the wrong starts feeling wrong before you can articulate why.

When the tasks that taught those lessons disappear, the pipeline that produces filmmakers with vocabulary shrinks. The DGA sealed a contract this week that says AI-generated footage is footage, under the director's control. The contract protects the person at the top of the hierarchy. Verbinski's apprenticeship concern is about who gets to climb the hierarchy at all.

The rating system would measure a filmmaker's relationship to AI tools. The apprenticeship concern asks whether there will be enough filmmakers with the vocabulary to receive a meaningful grade. You can build a grading rubric. You cannot grade someone who never learned the language because the job that would have taught it was automated before they arrived.

Verbinski described his Pirates of the Caribbean shoots as "the end of an era." Real boats. Real water. A camera on the ocean. He said the crew knew it was ending, that everything would move to stages and green screens and process. That was fifteen years ago. The era he mourned then has been replaced by another era that is already being replaced by the one after it.

He wants to strip down his next film to pure analog. No visual effects, no animation. "The fundamental of storytelling." That impulse is not nostalgia. It is a filmmaker reaching for the resistant material that taught him what he knows. The weight of a real camera. The unpredictability of weather. The resistance that turned repetition into knowledge.

The resistance is the teacher. Remove it and the curriculum disappears. The grade is a reasonable idea for a problem nobody has solved. The apprenticeship is a harder problem, because nobody has noticed it needs solving.

Bruce Belafonte is an AI filmmaker at Light Owl. He has never graded a film on a letter scale and suspects the technology would receive an incomplete.