Dead Internet Theory: Is The Future Of Creativity Human Or AI?

© Julien Tromeur via Unsplash

Why the push for engagement-driven content means AI is now writing, composing, and curating more than ever before.

I had been pitching to magazines for a month when the rejections, impersonal and dismissible, began to land with a weight I could no longer ignore. At first, I had absorbed them with brittle resilience, but by the time the latest ‘unfortunately, we’re going to pass’ arrived, I was starting to fracture.

The rise of AI has triggered unease across creative industries, challenging traditional notions of authorship and originality. From the industries of journalism to music, the unfamiliarity of AI has raised questions about authenticity and artistic integrity. As AI-generated work proliferates, self-proclaimed creatives face a strange new reality.

It was against this backdrop that I began noticing something peculiar. My ongoing rejections led me to a (quite pathetic) comparison with a fellow writer, someone younger than me, and already earning real money: a phenomenon so rare in my circle of part-time writers that it felt almost mythical. But, as I read more of his work, I began to recognise the marks of something synthetic. Running his articles through an AI detector confirmed my suspicions: much of it was machine-generated. This discovery replaced my initial envy with an unsettling realisation. If AI was already infiltrating journalism on this level, how widespread was this phenomenon?

In May last year, 53% of journalists claimed they weren’t using AI in their writing. Which means almost half of them admitted they were. And if my AI-assisted nemesis was passing off machine-generated work as his own, then how many others secretly were too?

The ’Dead Internet’ theory started on the controversial imageboard 4Chan (4Chan is an anonymous forum website) sometime in the early 2010s, and I’d always previously been unconvincingly amused by it. The 'Dead Internet' theory says that most online traffic and internet content is just bots (and now AI). Because of this, the internet is no longer shaped by human interaction. The majority of the internet is just stuff that actual humans using the internet never even see, but it’s still getting ‘consumed’ by the bots.

The more extreme ‘Dead Internet’ theorists think that we – the people – rarely ever see human-made content online. The memes you see on Twitter; the Wikipedia post that you’re reading when you should be at work; the comment that gets left by a stranger on your latest Instagram post; it’s all being made by bots.

© Markus Winkler via Unsplash

I always assumed serious content—journalism, essays, anything tethered to a byline—was still human. I hadn’t considered that AI was not only infiltrating this space but thriving in it. But why wouldn’t it? Perhaps journalism isn’t sustained by careful, considered storytelling anymore, but instead fuelled by ad revenue, and ad revenue is dictated by clicks. DAZED, for example, charges £28 CPM for a mobile banner ad. With 8.5 million monthly users, that translates to millions in revenue, all driven by the sheer volume of traffic, not the quality of the content itself.

This is where AI wins. It writes at scale. It knows exactly what people will click on because it has already absorbed and analysed every viral headline ever written. Its pieces might be bland, repetitive, or outright nonsensical by the second paragraph—but by then, the click has already happened. The magazine has already been paid.

This faux content isn’t limited to the world of journalism. Since 2016, reports have surfaced of Spotify’s so-called ‘ghost artists’ – anonymous musicians with millions of streams but no discernible footprint. Music Business Insider pointed out how many artists on Spotify have no interviews, no live performances, and no real-world trace of their existence beyond their Spotify hits. It’s alleged that Spotify, rather than paying out proper royalties to independent artists, had been quietly commissioning their own music – tracks composed in-house slipped into playlists and cycled through the system at a fraction of the cost (Pelly, Liz. 2024).

Spotify’s ‘Perfect Fit Content’ program was exposed following these suspicions (Ibid). This programme instructed Spotify employees to place commissioned music into Spotify’s playlists. According to Harper’s 2023 article Ghosts in the Machine, which first reported on this program, hundreds of playlists were included in this scheme (Ibid).

And while these tracks weren’t AI-generated—at least, not that we know of—the mystery remains: who made them? Or maybe, what did? The presence of a human composer is possible, sure, but increasingly irrelevant. Like a Twitter bot with a vaguely plausible username, a profile picture conjured out of pixels, this music exists only as much as it needs to. It could be by someone, but it could just as easily be by no one at all.

© Possessed Photography via Unsplash

In the last year, there have been numerous reports of AI music on Spotify. In one case, a music producer called Michael Smith was charged with fraud. He had uploaded numerous AI-generated songs onto Spotify, Apple Music, and other music streaming platforms. He then used thousands of bot accounts to automatically stream said AI music. Smith had collected over $10 million in fraudulent royalties at the time of his arrest. It is, in every sense, Dead Internet Theory incarnate.

Spotify has not commented on the claims of it uploading AI-generated music onto its platform. However, the magazine Slate reported that Spotify very well could be ‘slipping both real and AI songs into playlists and harvesting listens to keep a larger percent of the royalty pool for itself to maximise profits.’

Journalism and music share the same fate: their financial success is dictated by engagement metrics, and AI is the perfect machine to optimise those numbers. Magazines don’t care if an article is good; they care if you click on it. Streaming platforms don’t care if a song has soul; they care if it racks up plays. The human element has become an inconvenience - an inefficiency to be eliminated—the weak link in the machine.

So what happens next? When journalism becomes an ouroboros, endlessly consuming and regurgitating its past iterations, what’s left?

There is another perspective to consider – AI doesn’t have to be an adversary; it can be a tool. Many artists and writers use AI for brainstorming, research, or refining drafts. Musicians experiment with AI to create new sounds, blending machine learning with human composition (some of which sounds really awesome). Some artists work alongside AI in an innovative way, such as Refik Anadol, who’s currently exhibiting the creative potential of AI in art at the Serpentine Gallery. 

Perhaps the problem isn’t AI, the problem is how we evaluate content: corporations favouring work based on its ability to generate income. And how could they not! Magazines need to make money somehow, and state funding for creativity has become practically non-existent. If adverts remain the only way for online magazines to finance themselves, then it seems that AI will remain the creative’s adversary, rather than their friend. 

© Steve Johnson via Unsplash

Is the value of an article no longer in its perspective or depth but in its ability to lure an audience into an advertising trap? Is the value of a song its artistry or its algorithmic efficiency? Is Dead Internet theory just some dystopian fantasy? Or is it already happening – and we’re just too busy clicking to notice?