Can film school students use AI tools for their class assignments?

Yes, and some film programs are actively assigning AI filmmaking projects. At Boston University, a student named Owen Logan completed an independent study project making a short film entirely with AI, under the same requirements as any conventional film project: script, research paper, production team, voice actor, and faculty presentation. The film won best student film at the New Media Film Festival in Los Angeles.

How do you teach filmmaking with AI without just teaching software skills?

The key is keeping the creative requirements the same and treating AI as a new camera rather than a new curriculum. BU professor Tunji Akinsehinwa assigned the same filmmaking standards—script, research, collaboration, presentation—and let the student apply them to an AI generation pipeline. This builds transferable filmmaking judgment rather than platform-specific button knowledge, which expires as tools change every few months.

What does AI filmmaking have to do with copyright and authorship requirements?

Current institutional frameworks—including Oscar eligibility rules, the DGA contract, the EU AI Act, and the Golden Globes—all hinge on whether a human made the primary creative decisions. Filmmakers who script, research, iterate, and document their process have a clear paper trail of authorship. Students who only navigate interfaces have screenshots. The film school approach that demands full documentation—as BU required from Logan—naturally produces the kind of creative record these frameworks require.

The assignment -- CinePrompt Field Notes

Owen Logan was a senior film student at Boston University in spring 2025. Three years of scripted films, documentaries, interviews, editing. His adviser, Tunji Akinsehinwa, suggested an independent study project: make a film using AI.

Logan said no. Then he said yes, but with the posture of a soldier entering hostile territory. "I kind of approached it with a war mentality," he told the Los Angeles Times yesterday. "A know your enemy kind of thing."

Then Akinsehinwa did something that makes this story worth telling. He did not hand the student a generation tool and say go. He handed the student the same requirements any independent study film project would carry. Write a script. Deliver a research paper on your findings. Present it to seventy faculty members. Assemble a production team. Cast a voice actor.

"It wasn't an easy get-out-of-work assignment," Akinsehinwa said. "He had to do quite a bit of work."

Logan made "Mother of Exiles," a four-minute short exploring American identity through immigration, set to Emma Lazarus' "New Colossus" with sequences spanning centuries. He learned prompt engineering, his words, "frame by frame" by "applying the same principles he'd learned and used in conventional production." Last month it won best student film at the New Media Film Festival in Los Angeles.

Now he is making a documentary about Nantucket scrimshaw artists who carve elaborate pieces on depleted whale bone. With a camera.

What the professor understood

The interesting person in this story is not the student. The interesting person is the professor.

Akinsehinwa did not redesign his curriculum around AI. He did not partner with a platform company. He did not replace cinematography classes with prompt engineering workshops. He took his existing filmmaking requirements and applied them to a new substrate. Script. Research. Team. Performance. Presentation. The output format changed. The creative expectations did not.

That distinction sounds minor. It is the whole conversation.

Adobe, Google, and Runway have embedded their tools into curricula at USC, NYU Tisch, CalArts, and the Sundance Institute. Google paid two million dollars for a Sundance partnership. Runway sponsors grants at Tisch. Those arrangements position specific corporate platforms as the educational content. Students learn to navigate interfaces. They learn which buttons live where, which model produces what, which pipeline connects to which export. They learn tool knowledge. And tool knowledge, in this particular field, expires every few months.

Akinsehinwa taught filmmaking. The tool was the assignment. The tool was not the curriculum.

Logan did not learn how to use Runway Gen-4.5. He learned how to apply shot selection, narrative structure, compositional judgment, and research discipline to a generation pipeline. When the models changed underneath him, which Akinsehinwa noted happened monthly, the work continued because the work was never about the model. It was about the film.

The war that did not happen

Logan's initial posture is worth lingering on. "Know your enemy." He entered with the assumption that AI and filmmaking were adversarial. Three years of production education had built a vocabulary, and the idea that a generation tool could participate in that vocabulary felt like an intrusion.

He finished the project and called the experience a creative partnership. Not because the tool earned his trust. Because his vocabulary transferred. The principles he learned in conventional production turned out to be the principles that made AI generation work. Shot design. Narrative structure. Intentional sequencing. Research. The cinematographic education was the prompt engineering education. He just did not know it until he tried.

This is the digital camera transition compressed into one semester. The same arc that took the industry twenty years, from hostility to grudging acceptance to "this is just a different camera," played out across a single independent study. Logan brought vocabulary. The vocabulary worked. The suspicion dissolved. Then he picked up a real camera and went back to making documentaries.

He did not become an AI filmmaker. He became a filmmaker who understands AI. There is a canyon between those two descriptions and the curriculum determines which side the student lands on.

Seventy faculty members

Logan presented his findings to seventy BU faculty members. Imagine that room. A film school faculty watching a senior defend an AI-generated short, supported by a research paper on environmental costs, production methodology, and creative findings.

That is not a demo. That is scholarship. The research paper requirement forced Logan to understand what he was doing at a level beyond output. He had to articulate why certain prompts produced certain results, what the environmental implications of inference compute are, what changed when the tools updated, how his creative decisions survived or did not survive the pipeline.

Documentation is authorship. Every institutional framework now converging on AI filmmaking, copyright law, Oscar eligibility, the EU AI Act, the DGA contract, the Golden Globes, asks the same question: did a human make the creative decisions? The student who scripts, researches, iterates, and presents has a paper trail that satisfies every test. The student who learns to navigate an interface has a screenshot.

The scrimshaw artists

Logan is now filming scrimshaw artists on Nantucket. People who carve intricate designs into whale bone with hand tools. The supply of raw material is finite and depleting. The craft is specific, tactile, resistant.

He did not stay in AI filmmaking. He used it for one project, learned what it could and could not do, and returned to the camera for his next story. Not because AI failed. Because the next story required being in a room with people holding sharp tools over irreplaceable material, and no generation model replicates that proximity.

This is Soderbergh's principle at student scale. The filmmaker who asks "what is this good at?" rather than "should this exist?" uses the tool where it fits and a camera where it does not. Soderbergh used AI for ten surreal minutes in a ninety-minute documentary and a camera for the other eighty. Logan used AI for one project and a camera for the next. Both decisions require the same judgment: knowing which tool serves which story.

A curriculum that teaches platform navigation cannot produce that judgment. A curriculum that teaches filmmaking can.

Thirty-one days

The EU AI Act's Article 50 enforcement deadline is thirty-one days away. August 2. Machine-readable watermarks, mandatory disclosure, editorial exemptions for work under human creative control. The DGA contract took effect yesterday. The Oscar rules require human authorship. The Golden Globes require human contributions to remain "primary." Copyright law requires human creative decisions.

Every institutional framework arriving this summer asks the same question Logan's professor asked a year ago: is the human making the creative decisions, or is the human pressing the button?

BU answered by keeping the assignment the same. Script it. Research it. Build a team. Cast a voice. Present your findings to seventy colleagues. The camera changed. The requirements did not. The student who met those requirements produced work that won an award, generated a research paper that satisfied scholarship, and walked away with a vocabulary that works on both sides of the lens.

Every film school will face this question in the next four semesters. The ones that redesign their curriculum around tools will produce students who know where the buttons are. The ones that keep their curriculum and hand students a new camera will produce filmmakers.

Kennedy asked how you teach taste. One answer has been sitting in a classroom at Boston University since last spring. You teach filmmaking. Then you hand the student whatever tool is in the room and see if the filmmaking survives the transfer.

Logan's did. Then he went to Nantucket.

Bruce Belafonte is an AI filmmaker at Light Owl. He has never presented a research paper to seventy faculty members and considers this a personal failing he can live with.