The Discipline AI Didn’t Have to Earn

How generative AI is challenging the filmmaking judgment built through years of discovery.

I watched Jurassic Park again recently and it took me right back to being seven years old in a small theater in Wheaton, Illinois.

I remember the fear and excitement of it. The images. The sound. That T. rex coming to life on screen.

More than 30 years later, it still works. The computer-generated images hold because the story, performances, and practical effects are still believable. As a filmmaker, I can feel the discipline that went into bringing Michael Crichton’s story to life.

And this time, Ian Malcolm hit me differently.

Malcolm is brought to Jurassic Park to question what Hammond and his scientists have created:

“It didn’t require any discipline to attain it… your scientists were so preoccupied with whether or not they could that they didn’t stop to think if they should.”

He could just as easily have been talking about generative AI.

Every frame is like an artist’s brushstroke on a canvas. Humans have been bringing images to life for thousands of years — from cave walls in France to Georges Méliès sending audiences to the moon.

That process was discovery. Trial and error. Failure. Practice. Time.

Discipline.

Automation can help reduce dangerous human errors in medicine, aviation, engineering, automotive systems, and mathematics.

Art is different.

I don’t mean a technical mistake on set. I mean the creative mistake — the performance, camera move, or edit that wasn’t planned and becomes something unexpected and beautiful.

Trial and error develops judgment. The discipline teaches you when the accident is worth keeping.

That is what I am committed to understanding with Atlas Observe. More than that, I want to defend human authorship in filmmaking.

If AI can write, storyboard, generate, revise, select, and assemble most of a film, who owns the creative authority when the machine does 99% of the work?

Authorship has to mean more than writing a prompt and approving what comes back.

What did the filmmaker choose? Reject? Change? Protect? What did they discover and decide was worth keeping?

AI can make more of the film.

The filmmaker still needs to author it.

Boringly simple. Ridiculously serious about authorship and control.