For decades, filmmaking has been a strange combination of art and technology. Cameras became digital, editing moved from film strips to software, and visual effects evolved from physical models to computer-generated worlds. Every major technological shift changed how films were made without necessarily changing who made them.
Generative AI feels different.
It can create images, modify performances, generate environments, produce voices and increasingly turn a creative idea into something that looks close to finished footage. That makes AI more than another production tool. It potentially changes the economics of who needs to be on set, how much creative work is required and, most importantly, who owns the output.
My view is that the most interesting question is not whether AI will enter filmmaking. It already has. The real question is where filmmakers draw the line between using AI as a tool and allowing AI to become a substitute for human creative labour.
That distinction is likely to define the next phase of Hollywood.
AI Is Already Inside The Production Pipeline
The most important AI changes in filmmaking are not necessarily the ones audiences immediately notice. They are already appearing inside production workflows, from pre-visualisation and VFX to de-aging and localisation.
Hollywood has already tested what this looks like. The Irishman used extensive digital de-aging to recreate younger versions of its actors, while Indiana Jones and the Dial of Destiny digitally recreated a younger Harrison Ford. These examples predate the current generative-AI boom, but they established that digital manipulation could become part of mainstream filmmaking when it serves the story.
The backlash is not confined to major studios. The independent horror film Late Night with the Devil faced online criticism and boycott calls after viewers discovered that filmmakers had used generative AI for three brief interstitial images. The directors confirmed that they further edited the images and used them only briefly, but that did little to stop the controversy.
That reaction is revealing because it shows that AI adoption is not simply an operational decision. It can become a reputational one. Audiences may judge the use of AI not only by how much of a film it affects, but by whether they believe the technology replaced work that could reasonably have been done by a human artist.
That distinction matters for studios of every size. A production can use AI sparingly and still face criticism if viewers believe the technology crossed an unnecessary creative boundary. Hollywood therefore has to consider not just what AI can make cheaper or faster, but whether audiences will accept the way it was used.
For me, that is the real dividing line: AI should make filmmaking more capable, not make creative labour disposable.
Pre-visualisation Could Become AI’s Most Useful Creative Role
Pre-visualisation may be one of the areas where AI makes the most sense.
Before shooting, filmmakers need to imagine how scenes will look, where characters will stand, how a camera will move and how different visual elements will interact. Traditionally, this can require storyboards, concept artists, previs teams and sometimes physical or digital prototypes.
AI can dramatically accelerate that process.
A director could explore several versions of a scene within hours instead of waiting for each concept to be manually developed. A production team could test environments, lighting ideas or camera compositions before spending money on a physical set.
But faster does not automatically mean better.
One of my concerns with AI filmmaking is that the industry could become obsessed with the number of ideas it can generate rather than the quality of the ideas being selected. If everyone can create hundreds of visually impressive concepts, creative judgment becomes more valuable, not less.
That is why I don’t see AI making filmmaking talent irrelevant. I see it potentially shifting the premium from execution towards direction, taste and decision-making.
The filmmaker who knows exactly why a scene should look a certain way will still have an advantage over someone who simply knows how to generate attractive images.
VFX: The Real Risk Is the Collapse of the Career Ladder
The threat to VFX is not that artists will suddenly disappear from Hollywood. It is that the industry could gradually eliminate the junior work through which those artists become experienced professionals.
Rotoscoping, tracking, plate cleanup and background preparation may not be glamorous, but they have traditionally been the apprenticeship layer of digital filmmaking. Young artists learn production discipline, problem-solving and visual storytelling through this work before moving into specialised roles.
AI can automate parts of these processes and create immediate savings for studios. But if entry-level work shrinks too aggressively, Hollywood could eventually have plenty of senior supervisors and far fewer mid-level artists ready to replace them.
This makes the employment debate much bigger than “Will AI take jobs?” The real concern is whether studios are removing the first steps of the career ladder while still expecting a steady supply of experienced talent at the top.
The same problem extends to junior concept artists, assistant editors and entry-level writers.
These jobs are not simply labour costs; they are how creative professionals gain experience.
Hollywood could save money today by automating the apprenticeship stage and pay for it later through a weaker talent pipeline.
The Mid-Budget Film Is In a More Vulnerable Position
The AI debate usually focuses on two extremes: huge studio blockbusters and low-budget independent films. The more interesting economic risk sits between them.
A $150 million franchise film is unlikely to become fully automated. Its scale, stars, physical production and global distribution still require substantial human involvement. At the other end, AI could give independent filmmakers visual capabilities that previously required much larger budgets.
The $30 million to $50 million film faces a different problem. This has traditionally been Hollywood’s space for original stories that were too expensive for indie production but not commercially obvious enough to receive blockbuster budgets. If AI allows smaller productions to replicate some of the visual qualities of expensive films, studios could become even less willing to finance conventional mid-budget projects.
That could create a two-speed industry: increasingly expensive franchise spectacles at the top and inexpensive AI-assisted productions at the bottom, with the middle squeezed from both directions.
I see this as a bigger economic issue than simply asking whether AI lowers production costs. Cheaper filmmaking is not necessarily healthier filmmaking if it encourages studios to concentrate investment in franchises while reducing the range of stories being financed.
Your Face And Voice Are Becoming Production Assets
The most uncomfortable AI question in entertainment is not necessarily whether a machine can generate a background. It is whether a performer can be digitally reproduced without losing control over their own identity.
A digital replica can potentially recreate elements of an actor’s appearance or performance. Voice technology can also make it possible to produce speech that resembles a real person’s voice.
This turns a person’s identity into something that can potentially be treated like a production asset.
The issue became central during the 2023 SAG-AFTRA strike. Its eventual agreement established protections around digital replicas, including requirements for consent and compensation in defined circumstances. Producers generally have to obtain consent before creating and using a performer’s digital replica, while the agreement also establishes rules around digitally altered performances and synthetic performers.
To me, this is one of the most important developments in the entire AI debate.
A performer should not have to choose between working today and unknowingly giving away the ability to reproduce themselves tomorrow.
The concept of consent becomes particularly important because AI can make reproduction scalable. A traditional performance happens once, and filmmakers record it. A digital replica can potentially be reused across projects, scenes and formats.
That means the contract cannot simply ask, “Did you agree to AI?”
It needs to ask which AI to use, for what purpose, for how long, in which project and for what compensation?
Dubbing Shows The More Complicated Side of AI
AI dubbing and performance editing have already moved from theory into post-production. Flawless AI’s TrueSync technology, for example, has been used on Fall to alter actors’ lip movements after dialogue was changed, allowing explicit language to be replaced and a PG-13 rating to be secured without conventional reshoots.
For global streaming platforms, this capability has obvious commercial value. Dubbing is expensive and time-consuming, and AI can make it easier for films and shows to reach audiences in more languages. But when deep-learning systems modify an actor’s facial movements to match new dialogue, the technology is doing more than translating a performance. It is altering the performance itself.
This is especially relevant in markets such as India, where films routinely travel across languages and dubbing is an important part of how audiences experience characters. Humour, emotion, timing and cultural context have to survive the transition, not simply the words.
For me, the strongest model is therefore hybrid: AI can accelerate localisation, but human actors, translators and language specialists should retain meaningful creative control. The technology can make a performance travel further; it should not quietly redefine the performance without the performer.
The Writers’ Agreement Shows Why “AI vs Humans” Is The Wrong Debate
The 2023 WGA agreement provides an important example of how the industry can approach AI without pretending the technology does not exist.
Under the agreement, AI-generated material is not treated as literary material, companies cannot require writers to use AI, and writers must be informed if material provided to them contains AI-generated content. The agreement also preserves the WGA’s ability to argue that writers’ material may not legally be used to train AI systems. What I find particularly interesting is that the agreement does not simply say “AI is banned.”
Instead, it tries to establish boundaries around AI’s role in a human profession.
That is a much more realistic approach.
AI is unlikely to disappear from production. Trying to prohibit every use of it may simply push experimentation underground. The more useful question is who is responsible for the creative work, who receives credit, whether a worker can refuse AI use and whether their existing work can be absorbed into an AI system without permission.
In other words, the industry’s next fight may not be over whether AI is allowed.
It may be over the terms under which it is allowed.
Hollywood’s Real AI Battle Is About Bargaining Power
This is why the union battles matter so much.
The WGA’s fight focused heavily on writers’ control over writing and AI-generated material. SAG-AFTRA’s agreement addressed performers’ digital replicas and synthetic performers. These are different problems because writers, actors and other creative professionals do not contribute the same kind of intellectual property to a production.
But they share one underlying issue: bargaining power.
If an individual worker negotiates with a studio that has access to increasingly powerful AI systems, the balance of power can shift dramatically.
A performer might be told that scanning their face is simply another part of production. A writer might be handed AI-generated material and asked to “clean it up.” A VFX artist might be expected to supervise an AI system doing work that previously required a larger team.
Individually, each request can look minor.
Collectively, they can redefine an entire profession.
That is why union agreements could become some of the most important documents in the AI era. They are not just deciding how today’s technology should be used. They are establishing precedents for what creative workers can be asked to surrender in the future.
Hollywood’s AI Problem Is Also an IP Liability Trap
Copyright could become one of the biggest corporate constraints on AI adoption because Hollywood’s business model depends on owning exclusive intellectual property.
The U.S. Copyright Office has maintained that copyright protection requires human authorship. Its 2025 report concluded that purely AI-generated material is not protected by copyright, while human contributions to AI-assisted works can qualify depending on their nature and extent.
That creates a serious problem for studios. Imagine a company uses generative AI to create the visual identity of a new franchise character, creature or fictional world. If there is insufficient human authorship for copyright protection, the studio may have created something commercially valuable without obtaining the exclusivity that makes entertainment IP so profitable.
This is the IP liability trap: AI can reduce the cost of creating an asset while potentially making it harder to establish exclusive ownership over it.
That matters far beyond the film itself. Characters can generate merchandise, video games, licensing deals, sequels and theme-park revenue, all of which depend on controlling the underlying IP.
Hollywood could save on production today only to discover that it has weakened the legal ownership that makes the asset valuable tomorrow.
AI Could Make Content Cheaper Without Making It Better
There is another risk that receives less attention: AI could make entertainment so inexpensive to produce that the industry starts valuing volume over originality.
Generative tools can reduce the cost of visual development, localisation, background creation and other production tasks. That creates an obvious incentive for studios and streaming platforms to produce more content with smaller teams. But the economics of abundance are not necessarily favourable to creativity.
The danger is particularly significant for mid-budget filmmaking. If AI makes visually ambitious projects possible with smaller crews, studios may decide that producing more low-cost content is safer than financing original projects that require larger human teams. The result could be a market filled with technically polished material but fewer distinctive creative voices.
This is where human judgment becomes more valuable, not less. AI can generate variations quickly, but filmmaking still depends on deciding which idea deserves to survive. A writer’s understanding of character, an actor’s interpretation and an editor’s sense of rhythm are valuable because they involve judgment, not simply output.
As generative tools become widely available, technical execution will become less scarce. Creative judgment will become the differentiator.
What The Best AI-Powered Production Model Could Look Like?
I don’t think the winning model will be a film made entirely by humans or entirely by machines. It will probably be a hybrid.
AI could handle the parts of production where speed, scale and repetition matter most. Humans could remain responsible for storytelling, creative direction, emotional performance and final judgment.
A practical model could look something like this:
- Pre-production: AI helps generate concepts, storyboards, previs and production references.
- Production: Human performers and filmmakers remain at the centre of the creative process.
- Post-production: AI accelerates repetitive VFX, editing assistance, restoration and localisation.
- Dubbing: AI assists with translation and voice workflows, while human performers and language specialists retain meaningful control.
- Digital replicas: Explicit consent, defined usage rights and compensation become standard.
- Writing: AI can support brainstorming or research where permitted, but writers retain authorship and contractual protection.
- Final creative decisions: Humans remain accountable for what reaches the audience.
The key word here is accountability. If a studio uses AI to create a scene that is culturally insensitive, copies another artist’s style too closely or digitally reproduces a performer beyond the agreed terms, someone needs to be responsible.
A machine cannot negotiate a union contract. It cannot obtain meaningful consent. And it cannot take responsibility for the cultural impact of a film. A human needs to do it.
The Future of Film Is Not AI vs Humans
The biggest mistake would be to frame the future of entertainment as a battle between humans and machines. AI is already becoming part of production, and keeping it entirely outside filmmaking is neither realistic nor necessarily desirable.
The more important question is who captures the economic value AI creates. If a VFX system allows one artist to complete work that previously required a larger team, does the saving become higher studio margins, better compensation, more ambitious productions, or some combination? What if an actor’s digital replica is reused, does the performer share in that additional value? What If AI makes dubbing cheaper, do voice professionals participate in the new economics?
This is why the WGA and SAG-AFTRA agreements matter. They are early attempts to establish boundaries around authorship, consent, compensation and AI use.
My view is that AI should be used where it makes filmmaking faster, more accessible and more ambitious. But technological efficiency is not automatically creative progress.
The industry needs to distinguish between automating repetitive work and automating creative careers. If Hollywood gets that distinction right, AI could expand filmmaking. If it gets it wrong, it may simply make creative labour cheaper while weakening the ecosystem that produces great films.
