On Tuesday, Hyperwrite AI chief executive Matt Shumer wrote – or prompted, perhaps – a viral 8,000-word treatise about the state of his industry: Something Big is Happening.
Here’s the AI summary: we’ve seriously underestimated how much knowledge-based labour artificial intelligence can do better than we can – and the latest updates put us at a point of no return.
If yet another AI chief executive sounding off another alarm isn’t enough to trigger an existential crisis, then a generated clip depicting Tom Cruise and Brad Pitt in a fistfight on a roof should do it.
It was made in Seedance 2, a Chinese-built system from ByteDance, who also developed TikTok. According to filmmaker Ruairi Robinson, who created the clip, it was the result of a two-line prompt. Subsequent clips added near pitch-perfect dialogue, zombie ninjas and more.
In my mind, the clip is startling – though perhaps not for the reasons one might think. We’re only a few years removed from the near-universal mockery of AI-generated video. The early ones were wonky and unrealistic – they couldn’t quite get hands right, and bizarre artefacts would creep into the frame.
With time, they’ve become harder to distinguish, with only a few telltale signs that those very online can instantly detect. People with less screen time are often fooled completely.
With each passing update, however, those signs have been patched over. Through the past several months, I’ve found myself able to identify a generated clip only because the content is implausible – not because of the quality of the imagery. AI can now near-perfectly mimic mobile-phone footage, big-screen sheen and anything in between.
Despite what chief executive Shumer may imply, I don’t believe we’re approaching a point where AI becomes genuinely creative or capable of producing true art. It is trained on existing data. It recognises patterns. It recombines what it has already seen and produces an increasingly convincing facsimile.
In short, it can replicate style, but it cannot generate substance. It can approximate perspective, but it does not possess one.
That is not to say the technology won’t cause serious disruption. The average film credits about 500 crew members – a blockbuster closer to 1,000. How many of those roles disappear when the cost of performing technical work with a tool such as this approaches zero?
But there is another cost we are barely discussing. When we can generate clips starring the biggest actors in Hollywood that are indistinguishable from reality, something subtler begins to erode.
It’s a conversation I’ve been having with friends in the nature-documentary world – people who spent decades studying, travelling the world and waiting patiently to capture footage of rare animals doing extraordinary things. Some of those images – from National Geographic or BBC series – are now embedded in the cultural psyche.
Open TikTok, Instagram Reels or X today and your feed will be flooded with even more sensational clips. They look just as real. They are not.
You may know that consciously. But repetition has an unconscious effect. See something often enough and its rarity fades. Over time, images that once felt miraculous begin to feel ordinary.
If nature documentaries lose their sense of wonder, audiences drift. If audiences drift, funding dries up. And if funding dries up, the real work – the difficult, patient, human work – disappears with it.
Blockbuster filmmaking faces a similar challenge. Regulation is possibly needed to control the use of copyrighted material and likenesses. But as new systems emerge across jurisdictions, enforcement becomes harder. The floodgates are open – and they may prove impossible to close.
So what happens when our feeds are filled daily with anything we can imagine our favourite stars doing? If, next year, Pitt and Cruise actually teamed up for a major action film, would it still feel singular? Would it still feel worth leaving the house for?
On the press tour for the third Avatar film, director James Cameron repeatedly emphasised that what audiences were seeing was the result of tireless work by human animators. The fact that he felt compelled to say so is telling.
If something can be done effortlessly by a machine, does it still feel extraordinary when done by people? The danger may not be that AI replaces art. It may be that it dilutes our capacity to value it.
Spectacle used to be scarce. Scarcity gave it weight. When spectacle becomes infinite and frictionless, the question is not whether machines can make images. It is whether we will still care when humans do.
