I've been having a lot of conversations with people in film, music, and publishing lately. The mood is complicated. There's genuine excitement about new tools — and genuine fear about new economics.
Both feelings are right. Let me try to honour them both.
The tools that are helping
De-aging and restoration. Martin Scorsese used AI-assisted de-aging in The Irishman; Peter Jackson used AI to restore the Beatles footage in Get Back. These are craft applications — long, painstaking work that would have been impossible five years ago, now merely difficult.
Voice preservation for actors who can't speak. Val Kilmer used an AI model of his own voice, trained from archival recordings, to perform in Top Gun: Maverick after throat cancer surgery. The model was built with his consent, by his own voice-restoration company. That's the version of this technology that belongs in the world.
Subtitling and dubbing at pace. Streaming services are pushing content into dozens of languages. AI-assisted translation isn't replacing human translators — yet — but it is letting them work faster and across more titles than before. Good human editors still do the final pass.
Visual effects democratisation. A two-person film studio can now do things that required a twenty-person VFX house in 2015. This isn't taking work away from the big houses (the blockbusters still need them); it's creating work for small studios that couldn't compete before.
Music production assistance. Stem separation, mastering assistance, clever sample browsing, AI-driven session musicians for demos. Most of my friends in music production are using these tools, and most of them love them — as tools, not replacements.
The work being quietly taken
I want to be honest about the other side.
Background music and stock audio. AI-generated music has become good enough for many "music beds" — podcast intros, corporate videos, game backgrounds. This has directly reduced work for composers who used to make a living from library music. For the established composers, this is annoying. For the juniors who would have made a living from library music on their way up, this is career-changing.
Voice acting for non-starring roles. AI voice-cloning plus a good director is, for many commercial applications, indistinguishable from a hired voice actor. Game studios, audiobook producers, and ad agencies are using this. The voice actors' union has rightly made this a central bargaining issue.
Concept illustration. The mid-level illustrator who used to make a living producing variations of mood-board concepts for entertainment companies is finding that work drying up. The best ones are adapting — using AI as part of their own process, moving into art direction — but it's not a zero-sum shift for everyone.
Journalism. Local news has been in trouble for twenty years; AI-generated content is making it worse for the remaining small outlets that relied on aggregation and basic reporting.
Where I think the craft survives
Here's my honest belief, after a lot of thinking: AI is very good at average. It's very bad at idiosyncratic. The creators whose work has a distinct voice, a clear viewpoint, a specific reason for being — their work is, if anything, becoming more valuable.
What's getting hit is the middle: the competent-but-generic, the "could-have-been-anyone" work. That's sad, because that work is often where early-career creators found their footing.
The industry response has to involve not just technical solutions (watermarking, provenance, consent frameworks) but economic ones — new models for paying creators whose work trains the models, new pathways for emerging talent, new protections for the jobs that can still be done meaningfully by humans.
What I'd tell a studio or publisher today
- Lead with consent. If you're using AI trained on copyrighted work, fix that first. Either license the training data, or use models built on licensed data only. This is going to matter legally, and it matters morally.
- Keep humans in the credits. If a human was involved — writer, director, actor, illustrator — credit them. Don't hide the work that remains.
- Use AI to give your artists more time, not less. The best use of these tools is to remove grunt work from skilled people so they can do their best work more often.
- Pay attention to what your union says. The SAG-AFTRA, WGA, and similar agreements are not obstacles. They're early maps of a sustainable industry.
Why I'm still optimistic, carefully
Every previous technology wave in media — sound, colour, television, digital, the internet — was meant to destroy the industry. Each one created more jobs and more creativity than it destroyed, eventually, after painful transitions.
I think AI will too. But "eventually" is doing a lot of work there, and the painful transitions are painful for real people. Thoughtful studios can make those transitions gentler, and I think most of the ones I know are trying to.
If you're working through these questions — we'd love to help. Book a chat.
