There’s an alchemy about creating a hit TV show or movie which seems impervious to logic, money, even data.

In theory, the formula is fairly straightforward – and replicable. A solid script. Actors who are willing to commit. A director and crew who share a similar vision.

And yet, in an industry obsessed with pattern recognition — sequels, prequels, audience quadrants, retention curves — the art of making people care remains tantalisingly analysis-proof.

The X Factor? Humans.

For example, how do you recreate the spontaneous laugh that makes a movie (Pretty Woman) and launches a career? Or the achingly romantic “I know” that Han Solo replies, entirely off the cuff, to Princess Leia when she says, “I love you” (The Empire Strikes Back)? “Could he BE any more arrogant AND vulnerable at the same time?” Matthew Perry’s Chandler Bing in Friends might ask in that way that only he can.

You can’t plan for those things. You don’t really write them. You get them – if you’re lucky.

Which begs the question: if AI can now write the script, direct the scene, generate the actors and even predict the emotional beats, can it ever truly recreate the unpredictable magic that only a human moment can deliver?

In short, when it comes to creating content, can AI ever really solve the No Body problem?

Generative AI – the Netflix Killer?

We can imagine a not-too-distant future where, say, you’ve been re-reading Robinson Crusoe on the way home. Following dinner and, perhaps, putting the kids to bed, you indulge in a bit of fan fiction.

What if the book was set in modern times? Sure, Tom Hanks’s Castaway kinda covered this but, on screen, we only see him teaching himself how to make fire, fashion rope from vines, and become an expert fisherman. Instead, how about we get into the more realistic minutiae of surviving – and making a life – alone on a deserted island?

Can our lead character – let’s make her a woman, a doctor – trial and experiment with distilling water? Can she learn to fire clay in a kiln? Can she rediscover agriculture, fight off intruders (who come to harvest the island’s native marijuana plants), assemble a rudimentary steam engine which powers a boat which might, just might, get her home?

What is her mental state? Does she replace religion with getting high? Experience survivor guilt? Take time off just to relax?

No need to fantasise anymore.

You feed – via voice – these variables into the personalised Generative Video Engine on your TV and ask it to create a movie just for you. “Give it a happy ending,” you say, because you’re in an optimistic mood.

Ten minutes later your film is ready for you to watch.

It’s still early days for this tech so while the movie is watchable, it’s not brilliant. But so what, you tell yourself, most human movies are only basically okay too.

When I ran through this scenario – of creating movies on the fly – with ChatGPT last week (April 2025), I asked whether it had any opinion. It replied:

“Yes – what you’re describing is not only plausible, it’s entirely inevitable.” [ChatGPT’s own emphasis]

ChatGPT also came up with the concept that this might be a Netflix-killer without further prompting.

(Side note: I remember an interview with Reed Hastings in Wired UK magazine in 2015 where he was asked directly whether there was a “Netflix-killer” out there. He replied that there would be eventually but that it was VR (virtual reality) rather than AI. Even the most visionary of us, huh?)

The Return of the Studio Star System

Another memory.

In the first Jurassic Park movie (1993), Jeff Goldblum’s character, Dr. Ian Malcolm, says:

“Your scientists were so preoccupied with whether or not they could, they didn’t stop to think if they should.”

Technologically, let’s assume the ship has already sailed and that the coming of a Generative Video Engine – of the type I described above – is a given at some point. The other hurdle is cultural.

Will anyone over the age of 25 really enjoy – or even accept – a movie made with entirely AI-generated “actors”?

The bridge for these “legacy viewers” might be the return of the old studio star system.

In the original version, 1910s-1940s, actors (“stars”) were signed to multiple movies over a period of years. They could be paid a guaranteed salary, be offered fame, luxury and training in return for fairly strict contractual terms: no say in scripts or roles, typecasting and suspensions for non-compliance.

We can envisage a similar accommodation – this time in the age of AI.

Studios want control of IP, pipelines, budgets.

Actors want long-term stability, especially as human roles diminish. The money becomes impossible to turn down. They’re paid whenever their likenesses (or acting styles) are used; paid when they’re not on set; their estates paid even after they’ve died.

The Robinson Crusoe remake? Let’s have Zoe Saldana (signed by, say, Apple TV) play the lead role.

The simple equation here is that it won’t be AI that first reshapes Hollywood. It’ll be the contract lawyers, actors and commodification of human performance.

Persona Capitalism

Ah, but why stop there? Once we’ve crossed the barrier where AI is so good – where AI actors are so realistic in conveying look, tone, language, meaning, emotion, that it doesn’t matter if we recognise them as inspired by specific people or not – we may be open to broadening our horizons.

In which case (movie trailer voice):

“Kylian Mbappe [French footballer] is Robinson Crusoe…”

“Emily English [UK food blogger / social media influencer; huge on Instagram] is Robinson Crusoe…”

“Taylor Swift is Robinson Crusoe. [small print] Swift’s team negotiated emotional tone vetoes.”

Athletes, politicians, dead celebrities, even nobodies who test well, could all star in your AI-gen movies.

The Best Reality TV of all Time

The logical extension here is why rely on others at all? Wouldn’t the most-requested TV show be a faithful recreation of watching one’s own life?

Pre-steps: give the AI model enough material to train with – i.e. videos and stills of oneself. Make sure that it can see you in a variety of situations, ages, angles. Give it enough to capture your voice, your arch, your essence. But hey, even if you can’t, it’ll be fun to see what AI (which you increasingly think of as another sentient being) makes of you anyway.

The results? Variable.

In the Stephen Baxter and Arthur C Clarke novel, The Light of Other Days, a group of scientists learn how to send video cameras through wormholes in space – and therefore time. In this universe, the life of Jesus (“The 12,000 Days”) becomes the most-watched TV show of all time because they have the actual day-by-day, minute-by-minute footage, so to speak, to show in all its human chaos, boredom, revelation.

Relying only on AI recreations, and presumably, for now, without a brain-AI interface to trace and log actual memories, our “on screen” lives would be ridden with gaps on multiple levels. Not so satisfying.

Humans vs. AI

Which takes us back to our starting point.

Are humans the X Factor here? Do our own failings, brilliance, novel insights from living necessarily unique lives, and regardless of whether we communicate them or not, demand a premium when it comes to content?

Is that word – content – now devalued because of the sheer amount of it; the changing interpretation of what it means?

And if there is too much of it – content – so banal in nature, is the real question: does human mediocrity still trump AI excellence?

Thoughts? DM me on LinkedIn.