When Every AI Tool Looks Like a Layoff

Production Professionals Must Help Shape AI Before Its Role Is Decided for Them

Film and television workers are being asked to make judgments about technology that is still taking shape in front of us. New models, products, and uses are arriving faster than production practices, guild agreements, laws, and professional standards can adjust. People have deep expertise in AI research, labor, copyright, and individual crafts, though no one has a complete map of how these systems should operate across an entire production. The uncertainty leaves an opening for the people who understand the work to help shape what comes next.

In conversations with editors, producers, writers, composers, and other production professionals, I have often felt the discussion close down as soon as AI is mentioned. Finished-work generation, digital replicas, organizational tools, and systems designed to improve communication are swept under the same heading before their functions are examined. Many of the people stepping away from the subject possess exactly the knowledge required to draw meaningful boundaries around it.

Their reaction has a history. According to Bureau of Labor Statistics data reported by The Wall Street Journal, employment in Los Angeles County's motion picture industry fell from 142,000 at the end of 2022 to 100,000 at the end of 2024—a loss of 42,000 jobs in two years. A separate Otis College analysis found that by March 2025, only 26 percent of the jobs lost during the 2023 strikes had returned, while Los Angeles County shooting days in 2024 remained 42 percent below their 2022 level. In that environment, AI carries fears of fewer jobs, uncompensated training data, weakened authorship, digital replicas, and another round of promised “efficiencies” that can amount to fewer people being asked to do more.

Those dangers deserve direct resistance wherever they appear. Closing the entire conversation, however, leaves the design of tools, contracts, and production policies to studios, technology companies, and people who may have little experience doing the work. Adobe has already placed generative video tools inside Premiere Pro, while the WGA's 2026 agreement now requires companies to give the Guild written notice if they license writers' work to train commercial generative systems. AI decisions have already entered the software and contracts that shape creative work. Their timelines are set by product roadmaps, negotiations, and corporate strategy, far beyond the creative workforce's ability to pause them while it reaches consensus. Waiting outside the conversation leaves that work moving at the same speed, with less production knowledge in the room.

Even as development accelerates, the professional role of the technology remains unsettled. The problems it addresses, the authority it receives, and the people its efficiencies serve are still open to influence. That opening will narrow as products, contracts, and policies harden. Practitioners who enter now can still help determine where the technology belongs and where it has no place.

Every department holds knowledge that belongs in those decisions. A VFX artist understands where automation can relieve repetitive cleanup and where it begins replacing artistic choices or using someone's work without consent. An editor knows when faster access to a buried line reading or alternate take supports the cut, and when a system begins making the choices—the held look, the extra beat, the withheld reaction—that create performance, rhythm, and point of view. Writers know where organization supports their thinking, performers know when a production aid crosses into a digital replica, and craftspeople throughout a production know which parts of their process protect the quality of the work.

A television editor recently wrote to me after reading my essay about using AI to improve creative alignment before music is written. She was in the middle of a music pass and recognized the same communication failures in the work in front of her; imprecise direction had also sent her on wild-goose chases during other productions. At the same time, she was reading a warning from a sound editors guild about artificial intelligence taking jobs. She could see how the technology might help people clarify their intentions even as she was being reminded of its potential to eliminate the people doing the work.

Across more than twenty-five years composing for television, I have watched recording formats, editing systems, delivery methods, budgets, and schedules change. One problem has survived every technical shift: a note can name the wrong problem. Someone might say that a cue feels “too emotional.” The composer may remove the melody, thin the orchestration, flatten the harmony, reduce the performance, or pull the mix back. Each change follows the note while moving the music further from what the scene needs.

Several revisions later, everyone may discover that the cue entered ten seconds too early. The note identified emotion, though timing created the unwanted effect. Moving the entrance allows the actor to reach the moment first and can solve the scene with the original musical idea intact. The distinction is easy to miss without experience inside the process.

Chasing that note can involve a composer, assistant, music editor, and mixer. Each revision has to be created, conformed to picture, prepared, uploaded, and reviewed. The production generally pays the same fee while those practitioners absorb the communication failure through additional hours of work, often without additional compensation. Helping the original note become more precise could preserve that time without removing a single person or creative decision from the process.

That sequence contains two kinds of friction. The conversation that reveals timing as the real issue is productive friction—friction that is actually doing something. It challenges assumptions, clarifies intention, and can lead the group to a better idea. The unnecessary versions created before the problem is identified are wasted effort generated by an imprecise note.

From outside the workflow, the entire chain may look cumbersome: too many conversations, revisions, approvals, and handoffs. Practitioners can see which exchanges carry discovery and which failures only consume time. A useful tool needs to understand that difference before it begins removing steps.

This kind of pattern recognition is what experienced practitioners bring to AI design. It allows us to distinguish systems that generate finished creative work in order to reduce paid creative labor from tools that organize context, preserve decisions, expose ambiguity, or help people communicate more accurately. People who know the workflow can identify where assistance ends, where authorship becomes vulnerable, and where a promised efficiency would erase the conditions that allow good work to emerge.

Bruce Holsinger's novel Culpability gives this idea a more human shape. After questioning the desire for machines that are good for us or good instead of us, one character imagines a more demanding possibility: machines that “help us be better ourselves.” In production, that aspiration would appear in practitioners who are clearer in their intentions, more precise in their communication, and more deliberate in their judgment, with authorship still theirs.

CueMap™, the scene-oriented workflow I’ve been developing to help creative teams explore musical direction before a score is written, grew from my decision to enter that design conversation. The prototype turns scene context into a collaborator brief and an exploration prompt for a generative tool. Its scope ends before final scoring, leaving interpretation and creative decisions with the user. Building it has forced me to examine which communication problems a tool can reasonably address, what authority it should have, and how its limits might hold under real production pressure.

The technology remains early enough for many more practitioners to enter. Professional uses are still being invented, and expertise with a model or product offers only a partial understanding of its consequences across a production. Editors, writers, performers, VFX artists, composers, craftspeople, guilds, and production leaders each see parts of the work that technologists and executives may never encounter. Their knowledge can become product requirements, contractual protections, testing criteria, and enforceable limits.

That involvement has to begin while requirements, policies, and contractual terms can still change. Practitioners should identify the problems worth solving, distinguish wasted effort from productive friction, test failure modes, establish prohibited uses, and retain the authority to reject applications that threaten employment, consent, or authorship. Their expertise should be compensated, represented collectively where possible, and connected to decisions that can still change the outcome.

Film and television practitioners have protected story, performance, and craft through every technological change in how the work is made by bringing their judgment into the rooms where decisions happen. That judgment lives in the music pass where a vague note can send a composer, assistant, music editor, and mixer through another round of revisions; in the edit bay where a held look or withheld reaction changes the meaning of a scene; and in the writers' room where a scene's intention is still being worked out. New tools, contract language, and production policies are already bringing AI into those rooms. The people who understand what happens inside them need to be present while the technology's role is still being decided.

The rules are being written, so the people who understand the work need to help write them. Now.