“By 2005 or so, it will become clear that the Internet's impact on the economy has been no greater than the fax machine's." – Paul Krugman, economist.
People have been forecasting the death of the internet for as long as there’s been an internet. From Y2K, to targeted ads, to the erosion of net neutrality laws – there’s always one doomsday or another on the horizon for the internet. But the internet has always managed to roll on, unconcerned.
At the end of November 2022, OpenAI launched ChatGPT, 5 days later – it hit one million active users. By comparison, Instagram took two-and-a-half months to hit the same landmark. Netflix? Over three years. [1]
An explosive start in a world of explosive starts. Overnight it made its impact felt in every sphere of human activity, from art to education to industry. An undeniably transformative tool, a digital industrial revolution. Yadda yadda, you’ve heard it all before.
Yes, there were drawbacks – educators were suddenly inundated by essays penned by computers, difficult to correctly identify with the plagiarism detection software of the time. Publishers and magazines had to suspend submissions due to being flooded by ChatGPT-generated content.
But these were teething pains of an otherwise obviously transformative piece of technology. The equivalent of the death of Topsy the elephant relative to the invention of the electric lightbulb. Sad? Sure, but not exactly worth worrying about.
Code was being miraculously debugged, parking tickets overturned, cancer was getting identified early! A tool that promised to simultaneously end the drudgery associated with 21st century living and empower highly qualified professionals and school children alike; it could only be a good thing.
And yet…
Dead Internet Theory is a relatively new conspiracy. Emerging some time in 2021 (the specifics are relatively hard to pin down, with its origins being on forums and 4chan boards). Its proponents claimed that the internet’s activity was now mostly faked with bots, with previously bustling spaces now feeling like ‘dead malls’.
Whilst it initially gained relatively minimal traction outside niche internet communities, over the last couple of years, people have started to see it as - whilst not necessarily true - then perhaps a prescient warning of things to come.
As of 2024, Meta no longer reports daily active users, instead sharing the infinitely vaguer ‘family daily active people’, a metric which purportedly records registered users who visits at least one of their products (Facebook, WhatsApp, Instagram, Messenger etc.) once per day. But why?
Historically, obfuscation of key performance data is not typically a good sign, with perhaps the most infamous example being Enron’s deliberately incomprehensible quarterly reports where warnings of the company’s financial troubles were concealed in enigmatic metrics that looked good on first-parse. Enron’s downfall looked unthinkable until, suddenly, it was happening.
You only need to use Facebook for a minute or two to spot the trajectory the platform appears to be on. Countless AI-generated images of Jesus are shared by hourly-posting content farms. ChatGPT text above stable diffusion images of imaginary bed-bound people implore users for donations. The comments are invariably awash with more bot accounts with single-word replies; ‘AMEN!’, ‘AMAZING!’.
The issue is certainly not exclusive to Meta, with X’s (formerly Twitter’s) infamous bot problem having gotten much worse, despite repeated claims from Elon Musk that he was dealing with the problem. Google’s image search is littered with AI-generated results, Etsy’s store has gone from predominantly handmade to being jam-packed with Dall-E.
The mall is very busy, but don’t count the fingers on the other shoppers.
What does this all mean though? Grifters, spammers and bots are all nothing new. It’s very easy to see this as the just latest weapon in the escalating arms race for our attention. It used to be viruses, now it’s trucks covered in American flags.
The difference here, I believe, is one of structural faults – born out of a misalignment between company aims and desired user experience. Social media platforms may have been initially constructed to function as a digital town square, but advertising incentives have invariably shifted them into attention sinks, where their purpose is to maximise engagement with the app. As such, content moderation is often seen as an inconvenience for social networks, an expensive, manual process that often looks to prune the content that drives the most engagement.
A real: ‘Come for the chance catch up with your aunt Liz, stay for the rampant conspiracy theories.’ sort of vibe. This deliberate courtship with misinformation and extremism has placed the cracks in the dam that AI is now poised to burst through.
As big tech cuts jobs by the thousand, with those in content moderation amongst the worst hit, it’s starting to feel like we’re jettisoning life boats whilst simultaneously taking on water (to further mix my metaphors). These issues alone, feel like they could be course-corrected over time by natural market forces – however when coupled with the fact the digital landscape is built upon ad revenue, revenue currently being squeezed by bot traffic, hindered by a decrease in platform trust, artificially bloated user counts and easy-access tools for scamming all point towards a massive reckoning in the value businesses place on many forms of digital advertising.
This is all a massive downer, but what’s the relevance for people in business today? I initially wrote this with the intention of pivoting into a ‘but don’t worry, solutions are on the horizon, but we’ll need to act fast and be careful about how we implement them’ approach, but I don’t think that necessarily feels genuine to how I currently perceive the issue.
Yes, there are potential solutions, tools for detecting AI generated content are getting ever-more sophisticated. Yes, the internet is incredibly resilient, and capitalism is nothing if not responsive to when users’ needs are not being met.
However, I want the driving, if-you-only-take-home-one-thing message to be one of caution. We’re likely not seeing the beginning of the end for the internet, or anything even close to resembling that; but we could be seeing the end of the beginning. The big, exciting disruptors of our youth have at least partially calcified, and are perhaps overdue a disruption in turn, so when engaging with these titanic platforms – proceed with caution.
