AI vs Stardust: are Hollywood’s A-list at risk of digital replacement?
This piece first appeared on my newsletter, Future Proof. Subscribing to that helps support my work and means I can keep having ‘thoughts’. So, please subscribe.
In my younger (and more vulnerable) days, I was quite involved with the indie filmmaking scene in London. As well as shooting odds and sods of my own, I ran a networking event for aspirant filmmakers (and, indeed, am still the admin of its Facebook group, where people’s hopes and dreams continue to be broadcast, and vanquished, in public).
One of the things I gained from this experience was a sense of perspective. For me, this particular interest ran from about age 16 to maybe 22. It was formative but it was fundamentally unimportant. Aged 19, I made a ‘feature’ (in inverted commas because I think the runtime was actually about 70 minutes) film during the holidays after my first year of university. It was a lot of work, and it screened once at a cinema, and then I sort-of allowed it to fizzle. I hadn’t invested my life savings, I hadn’t spent decades trying to get it made. I’d produced it on a bit of a whim, and fuelled by youth and self-importance, I figured I’d do something else, something better, and could consign this one to the ash heap of Vimeo. (It is no longer available on Vimeo, but you can see the trailer here, if curious.)
But there were a lot of people, attending my events, who felt more invested in their success or failure. They might have been actors who’d quit their jobs, invested in headshots and were spending their weekdays auditioning for gormless film students (who only paid expenses). They might’ve been writers who’d been shopping the same script for a decade, increasingly desperate in their belief that it would be the next Sundance smash. Or they might’ve been directors whose experiences with university theatre groups made them believe that they had the right stuff to do a job on an episode of Vera or Line of Duty. For a lot of these people, success was a precious, chimerical commodity, not to be trifled with.
But almost all (including myself) were labouring under a fundamental misapprehension. The entertainment industry is not a combine harvester, which you can load up with diesel and then drive up and down a field until the riches of Britain’s agricultural bounty are yours. No, the entertainment industry is a magic carpet. And its fuel? Stardust.
Over the weekend, reports emerged of an actress called Tilly Norwood. Norwood is an elfin brunette with a playing age somewhere between 15 and 25. According to reports, Norwood is in the process of seeking representation (the act of looking for an agent) which is not especially noteworthy, except for one crucial fact: Tilly Norwood is an AI. Some within the acting community responded with gallows humour (“She was a nightmare to work with!!!!” deadpanned Lukas Gage) but most took this at face value. Mara Wilson, star of Matilda, condemned Norwood’s progenitors. “Shame on these people,” she wrote. “They have stolen the faces of hundreds of young women to make this AI ‘actress.’ They’re not creators. They’re identity thieves.” “Hope all actors repped by the agent that does this, drop their a$$,” opined Melissa Barrera. “How gross, read the room.”
Over and over again on Instagram, actors (serious actors, I might add) weighed in to condemn the existence of Tilly Norwood. Eventually, Norwood’s London-based Dutch creator, Eline Van der Velden, weighed in. “She is not a replacement for a human being, but a creative work,” Van der Velden wrote on Instagram. “I also believe AI characters should be judged as part of their own genre, on their own merits, rather than compared directly with human actors… I hope that we can welcome AI as part of the wider artistic family: one more way to express ourselves.” It was a very po-faced statement — fitting, given the very po-faced response to Tilly Norwood’s brief flirtation with the limelight.
The thing is, despite the sincere handwringing, Norwood’s existence is clearly a PR exercise for Van der Velden’s AI company, Particle6. If they are to go on to have success in AI-generated imagery for film and TV, it won’t be via the creation of ersatz celebrity, but through world-building and special effects. If Tilly Norwood’s ultimate aspiration were to become a background extra in a Hollywood blockbuster, I’d commend her realism. If it’s her ambition to replace Florence Pugh or Zendaya as a tentpole actress, I’d be worried for her cognition potential. There is undoubtedly a market for AI actors (as soon as they become cheaper than the SAG supporting artist rates), but not in lead roles. Because an AI actor headlining a movie can’t provide the magic carpet with what it needs to get off the ground.
I have a friend who made a movie. A low-budget movie, but one that got distribution through the major streaming services, and was shown on Sky here in the UK. It starred a few recognisable faces from British telly, but the limited budget was stretched quite evenly across the cast and crew. Several years on, his main takeaway is clear: he should’ve invested more of that money in ‘name talent’. He tells the story of taking the movie to the Cannes film market, where all the sales agents and distributors would happily take a meeting with him. When they sat down, they, without fail, opened with a simple question. “So, who’s in it?”
This is something that’s lost in the rush to produce AI slop. The question is not really about quality, but about quantity. For many years (decades even), the film industry has produced surplus films. More films than could possibly be watched, and certainly more than could ever make money. This is the painful lesson that all indie filmmakers eventually realise: ‘build it and they will come’, the famous maxim from Field of Dreams, certainly doesn’t apply to film. And so, the market has to seek a differentiator. In part, that’s the question of technical quality. Does the film look good enough for theatrical distribution? Or even for Netflix? But as the years have gone by, technology has bridged that gap and quality has diminished as a differentiator. So, what else is there?
There are other qualitative evaluations, of course. How good is the writing? How good is the acting? How good is the creative filmmaking? But (not to be a snob) one doesn’t have to scroll very far down the list of 2025’s top-grossing films (№1: A Minecraft Movie, №2: Lilo & Stitch, №3: Superman) to realise that these are not excessively important to a film’s commercial success. A film is far more likely to succeed because it has that-bloke-with-the-muscles-from-something-you’ve-already-seen than because it is a genuinely well-conceived piece of cinema. That is the stardust that powers the entertainment industry. It powers it because making a film is a problem that any idiot with a reasonable amount of money can solve, whereas marketing a film is a far knottier proposition. When someone asks you “so, who’s in it?”, what do you answer? Tilly Norwood?? Say what you like about Tilly, but she’s not going to do the late night shows.
And she’s never going to have the public profile required to go from curiosity to A-lister. I’ve just read
’s book Supremacy, about the race between OpenAI and DeepMind to create AGI. In it, she details the emerging parasocial relations between consumers and AI chatbots; relations that often seem to turn romantic. At the same time as I was reading this, I was also finding that Instagram’s algorithm kept serving me reels by an account called mayalanez_. Maya (is it ok if I call you Maya?) looks a bit like Tilly Norwood (isn’t it weird how all these avaricious AI thots look like teenaged white girls?). She’s a kind-of hybrid of all the e-girls who’ve made bank exploiting being attractive and young in the digital age. In some ways, Maya is the platonic (or not-so-platonic) ideal of this Gen Z dentistry advert. Except, of course, that she’s not real. “Your favorite virtual girl,” her bio reads (as well as revealing that she’s 20, apparently).
Maya Lanez has 318,000 followers on Instagram. ‘She’ utilises this following to redirect to her ‘spicy content’ on FanVue, a softcore alternative to OnlyFans. This is a business model exploiting masculine loneliness and the consensual suspension of disbelief that allows people to think they are in love with an AI chatbot. All the same, Maya’s 318k followers pale in comparison with the 8m who follow flesh-and-blood OnlyFans doyenne, Sophie Rain, or the 417m who follow the new Mrs Benny Blanco, Selena Gomez. The truth is that, even within a generation having less sex than ever and being more inclined towards falsely imbuing AI with consciousness, there isn’t the critical mass of lust required for AI to supersede real women. And Hollywood runs on lust; it is, after all, the biggest constituent part of stardust.
Tilly Norwood is a gimmick, and Maya Lanez a curiosity. Neither imperils conventional acting (pornographic or otherwise). But the reaction demonstrates something. Celebrity makes acting a powerful union within the creative industries. The response to Norwood’s existence has been more vociferous, more condemnatory than feedback on the ACTUAL, REAL ways that AI is impacting the livelihoods of artisans within their industry. AI is replacing jobs already — whether that’s in VFX or supporting actors, or in development and strategy. Nobody is calling that out, because the only group within the sector with any sort of mouthpiece are… the actors. The front-of-camera ‘talent’. And they’re busy railing against the imaginary possibility that viewers are going to fall for the unfuckable Tilly Norwood.
And so it’s important to re-conceive AI slop as an assault on quantity, not quality. In most ways, AI will not prove qualitatively inferior to human created product (after all, I am demonstrably capable of creating my own slop). But in an industry already running a huge deficit of success, slop has a diluting factor. More films does not create more appetite for films; it just creates more films. And so, more films means more films failing. More filmmakers not achieving their dreams. And in that environment, the differentiator — the stardust — only becomes more important.
Far from threatening their livelihoods, Tilly Norwood and the all-singing, all-dancing troupe of AI performers increase the eternal currency of real celebrity. The value of Tinseltown’s leading men and women is increasing. It’s time, therefore, for celebrities to lend their support — their voices — to their disempowered colleagues, outside the limelight, for whom Norwood’s digital brethren provide a much more real, much more scary, threat to their livelihoods.
