Let's Talk About AI - Part 2: Setting the Stage - The framework I adopted before using AI on this project.

In the first post in this series, I talked about using AI while making HAPTIC. I also talked about the importance of deciding where your own boundaries belong.

So, where were mine?

Before I dive into the tools, shots, and the things that took considerably more work than simply generating a prompt, I want to explain the framework that I adopted to help guide my decisioning. It came down to what I felt a tool could best contribute, along with what needed to stay human.

KEEP THE CREATIVE RESPONSIBILITY

The story and screenplay for HAPTIC were mine and that part of the creative process will always remain mine for the stories I create. So were the decisions about what the film was trying to say and how it should feel. AI could help me solve a particular problem, but I remained responsible for deciding whether the result belonged in the film.

That responsibility includes rejecting things. An image can be impressive and still be wrong for the scene. A tool can produce an answer that sounds useful and still miss the point.

I had to evaluate the work, make corrections, and choose what to use. If something ended up in the finished film, I approved it. I don't get to blame the software for a creative decision I made.

PROTECT THE PERFORMANCE

This was one of my clearest boundaries: AI should not replace live actors or the human performance they bring to a film.

The performances in HAPTIC were filmed. Casting, rehearsal, direction, and the final performance choices remained human. Where I used generated environments and visual elements around an actor, the filmed performance was the basis of that work.

There is a distinction worth being precise about here. HAPTIC also contains fully generated shots and portions of sequences. One transition moves from the filmed performer to a generated distant view of a raft. That extends the sequence visually; it does not generate the actor's central performance. Saying that every frame was photographed would be inaccurate.

Protecting the actor's work meant keeping that distinction in mind throughout the process, including how I describe it afterward.

GIVE THE TOOL A SPECIFIC JOB

I used AI for defined problems: visual shot references in preproduction, individual effects elements in post, and supporting environments. I needed to know what I was asking it to contribute and where my own work would continue.

In the effects work, that often meant building separate plates, combining them with the photographed actor, and manually compositing the scene. Rotoscoping, color, lighting, blending, and the decisions that made those elements fit together still required my time and judgment.

There were also places where AI wasn't useful for what I wanted. The Monster Hands sequence, for example, was created through manual compositing with no AI. Having a tool available didn't create an obligation to use it.

That was a practical boundary as much as an ethical one. The job was to finish the scene in a way that served the film. The goal here was never to use AI from start to finish, but to help supplement where either I simply didn’t have the budget or needed a piece to complete the whole of the puzzle.

UNDERSTAND WHAT YOU'RE HANDING OVER

Using a platform also means making decisions about what material you put into it. I need to understand what its policies say about uploads, privacy, training, and rights before deciding whether I'm comfortable using it.

Those policies can change. They deserve another look when the work or the platform changes, especially when other people's work is involved. If the practices don't align with what I'm comfortable supporting, I can choose a different approach.

The same care applies to information coming back. When AI assists with research or organization around the film, I still have to verify the information before relying on it. A useful starting point can still (and often does) contain errors.

KEEP USING YOUR MORAL COMPASS

I care about getting my stories made. I also care about protecting the people who help me make them. Both were part of deciding how to use these tools on HAPTIC.

An independent budget creates real constraints. AI helped me create things I couldn't have achieved in the same way with the resources I had. That mattered. I was still responsible for the choices I made along the way.

Concerns about labor, consent, intellectual property, privacy, and environmental impact deserve equal amount of care. Using AI on this film doesn't settle those questions, and I don't think a filmmaker has to pretend otherwise to talk honestly about the work.

These were the boundaries I brought to HAPTIC. Someone else may draw theirs differently. What matters to me is being able to explain my choices, remain open to questioning them, and take responsibility for the finished film. I’m equally excited and scared for how AI can help filmmakers get their stories told. At the end of the day, I don’t think it’s a clear-cut right or wrong.

The next post gets into the first specific problem AI helped me solve: getting the images in my head into a form I could actually use to plan the shoot.