The Internet’s Not Dead, and I Can Prove It
Weaponizing Google reviews to demonstrate ChatGPT hasn't taken over (yet)
Most conspiracy theories don’t evolve much after their creation.
Just seven years after the Apollo 11 mission, former U.S. Navy officer Bill Kaysing published the first known claim that the moon landing was staged by the government. In 2002, Buzz Aldrin, the second person to walk on the moon, was confronted by a conspiracist pushing the same narrative. Aldrin punched him in the face.
And just this summer, Scarlett Johansson, Channing Tatum, and Woody Harrelson starred in Fly Me to the Moon, a rom-com winking at this same decades-old notion that the government somehow faked the Apollo 11 mission. The nonbelievers were finally vindicated — NASA helped make the movie!
By comparison, the tenets of the Dead Internet theory have shifted a great deal in just a few years, which is a major reason why I find this conspiracy theory so fascinating.
The Dead Internet theory (DIT) began as a creepypasta. Its specific origins are hard to pinpoint, but Know Your Meme cites a chan post from 2019 as the first conclusive use of the term. In its early days, the DIT was clearly fictional, with its supposed believers spreading outlandish claims about the U.S. government controlling Facebook and Google1 and killing off the living internet in 2017 so as to replace your digital connections with bots powered by artificial intelligence.
From the chans, the DIT migrated to YouTube. Like most terminally online people, this is where I first encountered the suggestion that the internet as we know it could be fake. I stumbled across a video outlining the DIT during the depths of COVID, was briefly intrigued, and clicked away when the narrator started conflating the practice of advertisers harvesting our data with the purported takeover of the internet by artificial intelligence.
We were still more than a year away from ChatGPT 3.0 being released to the public, and I was content to consign the DIT to the dustbin of weird COVID cultural moments like the Gamestop stock rally and football games played to near-empty stadiums.
Fast forward a few years, and all of the sudden Shrimp Jesus has taken over Facebook.
As excellent reporting from outlets like 404 Media and Garbage Day shows, Facebook is teetering on the edge of becoming the most prominent example of the dead internet. Most of its viral newsfeed content is created by bots for other bots to like and comment on. The most charitable view of the social media behomoth is as a dead mall, with Messenger and Marketplace playing the role of Cinemark and Macy’s, the desperate anchors tied together by cavernous walkways lined by empty storefronts.
The rapid proliferation of AI has changed the DIT in a way few conspiracy theories ever do.2 These new tools have taken the DIT out of the realm of fantasy and placed it squarely in the real world.
And so the core truths of the DIT shifted. Proponents downplayed its more outlandish claims, like the assertion the internet isn’t just made up of bots, but that these bots are designed by the government to control us, an old-school conspriacy theory in the same vein of long-disproven notions about chemtrails and fluoride.
Instead, on podcasts and YouTube and TikTok, believers pushed the suddenly very real possibility that the majority of our interactions online are now with bots rather than people.
With legacy media outlets publishing articles written by AI under fake bylines, Twitter users resorting to nonsensical replies to test if a suspicious account was a bot, and even scholarly journals containing evidence of articles written by large language models, it was fair to wonder if the DIT had done the unthinkable and moved from theory to reality.
I don’t think social media has reached critical DIT mass yet. But, like many conspiracy theories, I believe that the DIT contains kernels of truth. As AI content generation continues to eveolve and becomes more accessible, the probability that an account I’m arguing with on Reddit is run entirely by a machine grows higher.
Much as my grandparents likely thought the idea of sending a man to the moon was absurd when they were kids, only to watch the moon landing live on television, I am convinced that I will one day witness a robot or software pass the Turing test. In fact, far smarter minds than mine disagree about whether this has already happened.
However, there are specific corners of the internet that convince me over and over again that humans still power much of what we see and likely will for years to come. One such example are the Google reviews scattered across Google Maps.
Fake reviews on sites like Google, Yelp, and Amazon have been a scourge since long before ChatGPT went public, which might make the reviews of Google Maps a surprising example to hold up as ironclad evidence of the web’s enduring humanity. But the more I stumble across the bizarre, inane, and downright nonsensical reviews people write on Google, the less I believe a Terminator future is imminent.
Only people could come up with such imaginative combinations of words as “I miss Afghanistan,” (Amilcar, one-star review, Richmond Bridge) and “was told by an old hag I couldn’t park by the library even to use the public restroom and then go” (Anders Luthersson, two-star review, Stinson Beach) that you will find littered across Google Maps.
My belief in humanity is bolstered not by the trolling one-star reviews of undeniably beautiful sites like Yosemite and Yellowstone — this fad has been exhaustively documented — but rather by oddly specific reviews of places and infrastructure that doesn’t usually merit any observations at all.
ChatGPT would never deign to write “unusually unpleasant to smell” in a review of the East Bay Municipal Utility District Wastewater Treatment Plant, because ChatGPT does its best to approximate typical human behavior, and it is highly atypical to take time out of your day to review the local wastewater treatment facility. But brian kong [sic] left those words of wisdom for the world to enjoy in a five-star review published in February, and for that I have no doubt that brian’s account is human.
ChatGPT and the other AI platforms give me a lot of reasons to worry. I currently work in communications, a field whose workers executives continue to insist can be replaced by software. I am strongly considering going to law school, and law is another profession rife with ChatGPT scandals. Effective accelerationists would love nothing more than to innovate me and millions of others right out of a job.
But as long as I can find pearls of wisdom like “I guess my only input is that it looks MUCH better with visible water than it did during drought” (delirium ? [sic], three-star review, California Aqueduct) on Google Maps, I will rest easy believing that online, humans reign supreme.
Things I Recommend This Week
The Fight Against Public Charters is Heating Up | Aviation International News
booktok, brainrot, and why it’s okay to be a hater | alisha not alihsha (YouTube)
I Never Expected To Run For Office—Here's What I Learned |
(Substack)What If Money Expired? | Noema
The Unofficial Gag Order of Jamil Al-Amin | The Root
Besides Google Maps, there are a few other corners of the internet I’ve trawled for examples of mankind's unparalleled ability to post, such as the Trivia section of IMDB. I’m thinking about compiling my favorite examples andsending them out to you, my subscribers — this time without any of the commentary about the DIT or ChatGPT.
But before I do this, I want to know if such a compilation would be something you’d enjoy. Let me know:
To be clear, these compilations would not become the new focus of IBT, but more like special bonus editions sent out in addition to my semi-weekly writings.
And regardless of how you feel about seeing more of these, thanks for reading today’s edition!
And not in the way Edward Snowden proved.
Again credit to Snowden, who proved the government really is listening to us.