I hit play and recognized two names before the first frame resolved: Nicolas Winding Refn and Hideo Kojima. Seconds later the image warped into something wrong — my favorite auteur turned into an uncanny stand-in. I felt protective, and a little furious.
I’m writing to you because you probably care about creators you trust, and because you deserve better than this.
See You in New York@Prada With@Kojima_Hideo @kuhlandhan @camsugarmusic pic.twitter.com/HrClMNocqB
— byNWR Official (@NicolasWR) May 26, 2026
At Prada’s Tribeca preview, a short film starring Kojima and Refn arrived — and it reads like AI-made footage
I watched Satellites II because my allegiance to creators is stubborn. Kojima shaped my youth: Metal Gear Solid, the tattoo on my shoulder that still makes me smile. That history gave me trust, and trust makes disappointment sharp.
On screen, the tech that made the film was the story
The project is a Prada collaboration for a Tribeca exhibition about love, language, and creativity. Instead of feeling like a conversation between two artists, the short reads as a demo reel for generative tools — and not a flattering one. The faces stutter. Lighting mismatches. Movement feels off, like a bad VHS tape.
Is Hideo Kojima using AI in his films?
Yes. Kojima has long flirted with new tech — his games fold AI into their plots, and he’s openly Hollywood-obsessed. That makes his adoption of generative models unsurprising. What’s worth asking is how those models were used: a research toy, a production shortcut, or a deliberate aesthetic choice?
I remember seeing Del Toro refuse this route — and that matters
Guillermo Del Toro publicly said last year he would “rather die” than use generative AI the way it’s often used. That stance frames this differently for me: two creators I admire chose opposite responses to the same tools. One said no; the other leaned in. You can judge the result for yourself.
Why does the AI Kojima look so unsettling?
The uncanny valley isn’t a glitch in feeling; it’s a clash between method and empathy. Generative models trained on scans and datasets can reproduce features, but they can’t reproduce history, intent, or the tiny habitual ticks that make a face live. The outcome here reads like a celebrity mask run through an algorithm — its face felt like a wax museum knocked over.
The industry context: east vs. west and the creative cost-benefit
In Japan and parts of Asia, creators and studios experiment publicly with synthesis tools more readily than many Western houses. Kojima’s aesthetic has always courted the new; for him, tools and themes overlap. You should still weigh what you gain — faster iteration, novel imagery — against what you might lose: texture, human nuance, trust.
Platforms and tools matter here: people are using Midjourney, Stable Diffusion, Runway, and custom pipelines from studios. Each chain introduces its own artifacts. When you see a celebrity likeness produced by one of those stacks, judge both author and assembly line.
The emotional cost is real, not hypothetical
When a creator you revere releases something that looks half-made, it feels personal. I’m not accusing Kojima of selling out; I’m cataloging the fallout: fans stunned, industry watchers shrugging, and a handful of artists doubling down on analog methods. For me, that trust gap is why this hurts.
I care about artists experimenting. I also care about craft. Watching a beloved creator hand their presence over to generative pipelines without a clear artistic logic feels like seeing a signature photocopied. The film didn’t read as a choice so much as a convenience, and that stings.
You should expect better from icons. If a name matters to you, demand clarity: what was made by the person, what by the tool, and why that split matters.
For those keeping score: this isn’t the end of Kojima, and it might not even be a misstep if framed differently. But if this becomes the pattern — famous faces layered over generic generative output — then yes: reputations get thinner with every instance. Like a short story read aloud with every other line redacted, the message fragments.
I want to hear what you think. Are you willing to follow a creator who hands their likeness to an algorithm, or do you draw a line at human presence?