- The Tynan Files
- Posts
- The Internet is drowning in AI spam
The Internet is drowning in AI spam
The web is getting worse, and Google deserves much of the blame
‘Ummm, spam,’ said the robot. Source: Midjourney.
Remember when the Internet was cool? Those were the days.
Multiple polls and studies reveal what many of us instinctively feel: The Internet (and more specifically Google) is getting shittier by the day. And while AI is not to blame for creating the flaming pile of rodent excrement the net has become, it's shoveling even more feces onto the fire, then dousing it with kerosene.
A January 2024 study by researchers at Amazon Web Services estimated that nearly 60 percent of all sentences published on the web have been written by robots, a figure that's projected to hit 90 percent by 2026. At this rate, it will happen even sooner.
Ever since it became clear that a tsunami of combustible AI-generated crap is about to come crashing down upon us, Google has been working overtime trying to keep AI spam out of its search results. So far, it has failed miserably.
ICYMI: Welcome to the Enshittocene
A study recently published by Originality.ai found that the amount of AI spam contained in top 20 search results on Google rose from 2.4 percent in 2019 to just over 11 percent today. And while some of that AI-generated content is legit, much of it isn't.
As Originality CEO and founder Jon Gillham told me recently, "Not all AI content is spam, but all spam is AI generated."
Because why pay a bunch of slumdog wallahs in Hyderabad 5 cents an hour to churn out meaningless-but-search engine-friendly gibberish when ChatGPT and its robo-friends will do it for free?
Original sins
Originality.ai is one of a handful of AI tools designed to detect material generated by other AI tools. (The ongoing tragedy of robot-on-robot violence.) Though it's hardly mistake free, Originality is generally considered the best of the lot.
ICYMI: Did a robot write this post?
Gillham started out in the content marketing business. The idea for Originality came out of concerns his clients had that their websites would get body slammed by Google for using AI content, so they needed some way of finding out if their writers were using AI tools to generate the text.
A PR person sent me the doc scanned here, pretending it was written by her client.
And in fact, some of Gillham's writers were using Jasper, an AI-based writing tool – kind of a grammar checker/writing prompt on steroids. It's a perfectly legitimate use of AI, but one that could land a site on Google's shit list. And nobody wants to be on Google's shit list. So they use tools like Originality to determine how AI-ish something looks before they hit the publish button.
Some people also use the tool to check whether students are using ChatGPT et al to do their homework for them — something that Originality is not designed to do, Gillham notes; the tool is purpose-built to analyze web content. And sometimes it flags human-written content as machine generated.
Source: Originality.ai
False positives are unavoidable, he admits. Gillham says the false positive rate is under 2 percent, which isn't bad, but still not very helpful if you're a human writer accused of impersonating a machine. To mitigate that, Originality offers a free Chrome browser extension that literally watches you as you write (assuming you're using Google Docs) so can prove it was you who came up with that witty bon mot, not Robby the Robot Writer.
It's not clear what tools Google is using to detect machine-generated spam. But clearly they aren't working.
The spam also rises
Why do people create these spammy sites? For exactly the reason you think: money, honey. The higher you rank in Google searches, the more web traffic you attract, the more people click on affiliate links in your crappy AI-generated content, and the more dosh you take home. [1]
Do a search for any product and I guarantee you'll see bogus "top ten reviews" sites generated by AI that pretend to evaluate products and make personal recommendations. They're crowding out legitimate publications that pay actual human experts to evaluate this stuff. And they make it increasingly harder to tell real reviews from robot sludge; in the end, everything tastes like spam.
Nilay Patel, editor of The Verge, used Google Gemini to create a bogus printer review to illustrate how this works.
Source: The Verge.
Ever since Google emerged from the womb in 1997 and began to devour the world, people have devoted themselves to outsmarting it, often in sleazy and underhanded ways. Many have made a metric shit ton of money in the process. This is why some truly lousy sites rise to the top of Google search results while more worthy publications sink into obscurity.
The Verge published an outstanding article on the black arts of search engine optimization (SEO) last November titled "The people who ruined the Internet," and what they’ve left in their wake.
Author Amanda Chicago Lewis writes about the
"... universal, non-Google-specific resentment and rage about how the internet has made our lives so much worse in so many ways, dividing us and deceiving us and provoking us and making us sadder and lonelier. Decades of American optimism about the wonderful potential of technology, from the Moon landing to personal computers to the iPhone, had finally, in the last few years, broken down into comprehensive chagrin at the petty, pathetic, and violent world enabled by our devices."
Preach sister, preach.
Google uber alles
Google and SEO are not entirely responsible for the raging rat feces fire we've inherited, but they definitely own a big chunk of it.
The impact of Google's dominance over digital media is well documented. Roughly 26 cents out of every advertising dollar earned on the Internet goes right into Google's pockets (Meta/Facebook and Amazon absorb another 20 and 14 percent, respectively). That leaves less than 40 percent for the websites that are creating the content these services point you to.
Turns out that building a system that helps you find information on the Internet is way more profitable than generating the information in the first place. And it explains why independent digital media is on a respirator in the ICU, while the spammers and the bots are jumping up and down on the air hose.
I wish I could tell you it's going to get better. Instead, what will happen is that search engines will be replaced by AI assistants that simply answer any questions you might have, without sending you to the websites where they found the answers. That will result in less traffic going to spammy sites — but also fewer clicks on legitimate ones.
The AI Overview to the question, "Is Google evil?"
It's already happening. Google just rolled out what it calls "AI Overviews," which sit on top of search results and relieve your mouse fingers of the burden of clicking. Which ultimately means even more digital ad dollars going into the pockets of Big G.
I don’t think this is what God and Al Gore intended when they invented the Internet.
Is it time to break up Google? Share your thoughts in the comments below, or email me: [email protected].
[1] Affiliate marketing works like this: Every time someone clicks an affiliate link in an article and goes on to buy the product in question, the publication takes a small percentage of the sale. It's a legit business model that was perfected by The Wirecutter and is now used by practically everybody.
Reply