I know I’m real. And you, dear reader, know you’re real. But do you ever suspect that everyone else on the internet is acting strange? That the spaces you used to frequent feel a bit … dead? You aren’t alone. “Dead internet theory” first hit the web almost three years ago, propelled to the mainstream by an essay in the Atlantic by Kaitlyn Tiffany:
Dead-internet theory suggests that the internet has been almost entirely taken over by artificial intelligence. Like lots of other online conspiracy theories, the audience for this one is growing because of discussion led by a mix of true believers, sarcastic trolls and idly curious lovers of chitchat … But unlike lots of other online conspiracy theories, this one has a morsel of truth to it. Person or bot: Does it really matter?
At the time it was writing, the deadest part of the internet was Twitter in its moribund pre-Musk years. The site’s aggressive curation served up the same “relatable content” to hundreds of thousands of users, who made tweaks to posts like “i hate texting come over and cuddle me” and then reposted them. The distinction between person and bot was also being blurred by a recommendation algorithm that drove people to act like bots.
Beyond that central idea, the 2021 version of the conspiracy theory took some weird turns. One proponent, Tiffany wrote, “suggests that the internet died in 2016 or early 2017, and that now it is ‘empty and devoid of people,’ as well as ‘entirely sterile’… As evidence, IlluminatiPirate offers, ‘I’ve seen the same threads, the same pics, and the same replies reposted over and over across the years.’”
The theory wasn’t wrong – it was just too soon. Talking about a dead internet the summer before the release of ChatGPT is like the Guardian colleague who confidently declared, in the summer of 2016: “It’s been a mad rush of political news since the 2014 local elections, so it’s weird to think that there’s just Brexit and the US election and then everything will be quiet for the next few years.”
In 2021, the internet felt dead because aggressive algorithmic curation was driving people to act like robots. In 2024, the opposite has happened: the robots are posting like people. Here are just a few examples:
-
On Twitter itself, after Musk rescued the site from the frying pan and tossed it in a volcano, an ill-conceived monetisation scheme has made it profitable to buy a blue checkmark, attach it to a large language model, and set it running wild replying to viral content. The social network now pays verified users a proportion of the ad revenue received from their own comment threads, turning the most viral posts on the site into a low-stakes all-bot battle royale.
-
Death permeates Google. The top of its search results is a valuable position – so valuable that businesses competing to be there have no spare money to actually write their articles. No problem: ChatGPT can churn something out in a second. Of course, that’s only valuable if the resultant visitors are humans who you can make money from. Bad news, because …
-
… across the web, bots account for around half of all internet traffic, according to research from cybersecurity firm Imperva. Almost one-third of all traffic is what the company calls “bad bots”, doing anything from ad fraud to brute force hacking attacks. But even the “good bots” are struggling to earn that categorisation: Google’s “crawler” was a welcome sight when it was updating your search entry, but less so when it was simply training an AI to repeat what you wrote without sending any users over.
-
And then there’s Crab Jesus. An unholy marriage of Facebook content farms, AI-generated imagery, and automated testing to work out what goes the most viral led to weeks of viral content featuring combinations of Jesus, crustaceans and female flight attendants. In one such image, Jesus was pictured eating shellfish wearing a jacket made of prawns. More confusing was the image of a sort of crab-centaur saviour walking along a beach arm-in-arm with what appears to be the entire crew from a long-haul flight. It was, at least, interestingly bizarre – a step up from the previous viral chum of the 122-year-old woman posing in front of her homemade birthday cake.
I want to provide a ray of hope here, a nice little tip for how to insert some vitality into the internet again, but I can’t. It really does feel as if the public-facing net is in the latter stages of a zombie apocalypse. The good news is that there are safe havens. “Private social”, like WhatsApp and Discord servers, can hide from the onslaught in secrecy, while smaller communities such as Bluesky and Mastodon are safe through obscurity, for now.
In the mid-term, I expect to see large platforms cotton on to the wasteland their services have become, and use a combination of account verification and AI detection to try and restore some humanity to their offerings. Whether it will be too late by then, though, is an open question.
Musk still needs his Twitter sitter
Elon Musk in Beijing in 2023. Photograph: Tingshu Wang/Reuters
There is, at least, one human still on the internet: Elon Musk, who is so addicted to posting that he spent $44bn to be called an idiot on a platform he owns. So his latest legal defeat will have hit him where it hurts, after the US supreme court declined to take up his plea to be freed from a court-appointed posting babysitter. From our story:
The supreme court on Monday rejected an appeal from Elon Musk over a settlement with securities regulators that requires him to get approval in advance of some tweets that relate to Tesla, the electric vehicle company he leads.
The justices did not comment in leaving in place lower-court rulings against Musk, who complained that the requirement amounts to “prior restraint” on his speech in violation of the first amendment. The ruling comes a day after he made an unannounced visit to China aimed at sealing a deal to roll out Tesla’s driver assistance features there.
For those without an encyclopaedic memory of all things Elon: in 2018, Musk tweeted that he had “funding secured” to take Tesla private. The company never went private, and in ensuing lawsuits it turned out that he had, at best, a few discussions about it. To put an end to the legislation, Musk agreed to resign as Tesla chairman, pay $20m and have an in-house lawyer pre-approve all his social media posts about the electric carmaker.
skip past newsletter promotion
Alex Hern’s weekly dive in to how technology is shaping our lives
Privacy Notice: Newsletters may contain info about charities, online ads, and content funded by outside parties. For more information see our Privacy Policy. We use Google reCaptcha to protect our website and the Google Privacy Policy and Terms of Service apply.
after newsletter promotion
He’s regretted it ever since, battling to have that part of the agreement (which he entered in to voluntarily, to avoid a damaging court case) overturned. “The pre-approval provision at issue continues to cast an unconstitutional chill over Mr Musk’s speech whenever he considers making public communications,” his lawyers argued.
Well, the supreme court in the US doesn’t care; it didn’t take his case, implicitly determining that no real constitutional issue is at stake.
The weird thing is that the in-house lawyer already seems to be taking a very hands-off approach to Musk’s posts. On Friday, he responded to an allegation from early Facebook employee Dustin Moskovitz that Tesla was “the next Enron” by posting a picture of a dog draping its testicles over the face of another dog. (Click at your own risk.) If that’s Musk tweeting with an “unconstitutional chill”, I’d hate to know what he’d send if he felt truly free.
The wider TechScape
The artwork for Everyone Knows That. Illustration: Getty; Guardian Design