Hundreds of businesses just killed their websites with AI

Since the ‘90s, the professional world has been rocked by technical innovation. From email and personal computers to Blackberries and eventually iPhones, each decade introduced a radical new tool to make work easier and more efficient. The latest in this string of office work innovations is generative AI.

From drafting email templates to generating images, video, and music, the rate at which generative AI is evolving is breathtaking—and a little scary. Within weeks of OpenAI’s public release of ChatGPT in 2022, the marketing world was inundated with AI tools focused on content creation. These tools made bold claims about their ability to beat checks for generated content, deliver SEO results, and generally revolutionize the world of content marketing.

It’s been almost 18 months since then. How’s the revolution going? According to the latest SEO research, not great.

Let’s take a look at AI-generated content, its impact on content marketing over the past year, and why AI may have just killed hundreds of websites. But first, a history lesson…

When the cure becomes the culprit

Aspirin was the miracle drug of the early twentieth century.

More tolerable and effective than previous painkillers, Bayer’s wonder drug became a cure applied to many illnesses and conditions. It reduced fevers, eased aches and pains, and caused fewer and less harmful side effects than pain medicines that came before it. Similarly to today’s opioid epidemic, doctors prescribed aspirin without fully understanding its effects and pharmacology. In the early twentieth century, It was not uncommon for patients to be prescribed grams of aspirin daily—up to a gram every three hours, or over twice the maximum daily dose now considered safe.

It’s been implicated as a cause for high mortality rates during the 1918-19 Spanish Flu. High doses of aspirin can cause respiratory complications. Add a vicious flu virus to the mix and this pain-killing fever-reducing wonder drug may have resulted in thousands of excess deaths.

Today, AI is like an overprescribed medicine killing patients across the world. In this case, the patients are websites.

An AI Overdose: How AI Blog Content Kills Websites

We warned of the dangers of AI content over a year ago in our article “Why Chat GPT is bad for SEO content generation.” AI content generation is appealing to marketers and business owners alike. Content writers are expensive; AI like ChatGPT can produce serviceable content for free.

ChatGPT is just one tool of many. While it’s the most commonly known, it’s not the primary culprit for this AI overdose crisis. More specialized tools exist that generate content at scale. Configure topics, select keywords, develop a prompt, and hit “go.” Within a day, an AI content architect could spin up hundreds of posts.

There’s compelling evidence that some publishers employ AI tools that scrape the web—particularly sites like Reddit and Twitter—to generate articles on trending topics without any human input. The results can sometimes be laughable, like when World of Warcraft players trolled publishers by talking up a fictional new feature called “Glorbo” on Reddit. The trolling worked; AI content farms mistakenly produced articles hyping up Glorbo, an ill-defined feature that does not and will never exist.

AI is a like a toddler

Young children learn through mimicry. Monkey see, monkey do. The understanding part comes later. In this way, AI is like a perpetual toddler. Fundamentally unable to understand what it produces, or the prompts it’s given, generative AI mimics whatever it’s been fed as closely as possible. In our previous article on AI content generation, we highlighted how AI cannot produce unique ideas or perspectives, arguing this would be one way that AI content would fail to produce SEO results. Google cares about Expertise, Authority, and Trust. Toddlers typically lack in all three areas.

Google knows AI content is a problem

In the SEO world, there have been rumblings of a major shift in how Google handles AI content for some time. Right now, Google benefits from being perceived as the only game in town. Its internet search market share percentage is in the high 90s. But Google—and every other search engine—has a problem. Their primary product is getting shittier.

It’s debatable when “peak Google” occurred—the high-water mark in search results quality. Early Google Search was excellent compared to the mediocre offers on the market, but it was rudimentary compared to today’s index. We can be certain that today is not “peak Google.” Dozens of articles across the web highlight how Google search quality is in decline. You’ve probably experienced that decline yourself. How often do you search for something only to be met with a first page full of affiliate link spammers and annoyingly insincere top 10 lists? Why does Forbes rank on the first page for everything from medical advice to “best bluetooth headphones for 2024?”

The reasons for Google’s search quality woes are complex. SEOs are to blame (sorry) for over-optimizing website content until it satisfies search engines over readers. The biggest brands and most profitable websites are the ones with the budget to run SEO campaigns to maintain or improve their rankings, but these giant websites aren’t always the ones that offer the best result.

AI sure hasn’t helped. A powerful and effective search engine cuts through the noise to find quality results. AI content generates exponentially more noise than ever before. One of Google’s major issues over the past year is that AI content generation for SEO worked. In too many cases, clearly templated or generated articles outrank legitimate, human written content.

Google knows it, so they’re doing something about it.

The March Algorithm Update: A Killing Blow for AI SEO?

In March 2024, Google dropped a major update. Some SEO experts believe it’s the biggest and most disruptive update in five years, on par with Google’s biggest updates ever. The update targets websites that employ a few low-quality or spammy tactics. While Google’s official release about the March update doesn’t exclusively focus on AI content—in fact, the phrase “AI” never appears in their consumer-facing blog post about it—it’s alluded to with references to content automation and low-quality experiences.

It can take months for the full picture to emerge after a major update like this one. What’s already clear is that the worst abusers of AI content generation felt the brunt of this update.

Early results point to catastrophic changes for websites over-reliant on AI content. A study by Originality.AI found that hundreds of business websites have been delisted from Google since March 1. Of them, 100% used AI content. 50% used AI content for 90-100% of their posts and pages.

There are many thousands if not millions of websites still employing spammy AI tactics whose listings have not been removed. Hopefully, their webmasters are quaking in their boots. Google’s March update puts AI tricksters on notice: deliver value with your content or expect to face the consequences.

Should you be worried about your B2B website?

Fortunately, most people—including those who may have used AI content for some of their blog posts or pages—shouldn’t panic. The websites bearing the brunt of the blow are those with egregious content issues. Some of the delisted websites posted hundreds of articles a day, all of them AI generated. If your small business website has a few AI blog posts, it’s likely you won’t be significantly impacted.

With that said, it may be a good idea to remove AI content from your website. Despite our reservations, we’ve dabbled with AI content generation ourselves. The outcome, in most cases, was an article that didn’t perform. Even if it got clicks, it didn’t get engagement. Use tools like Google Analytics and Google Search Console to verify how your AI content performs. If it’s not performing exceptionally well, consider heavily revising or even deleting the post. Why expose yourself to risk if there’s no associated reward?

Will this update make Google un-shitty?

This is the burning question for both Google and its users. Google’s usefulness as a search engine is a primary revenue driver. While it’s hard to imagine another search engine dethroning Google any time soon, they can’t ignore the rising tide of bad content polluting their results.

In its blog post concerning the update, Google claims they will see a 40% reduction in low quality content in their results—kind of. Here’s the exact quote:

Based on our evaluations, we expect that the combination of this update and our previous efforts will collectively reduce low-quality, unoriginal content in search results by 40%.

When we break it down, they’re saying that the March update plus an unspecified number of previous updates over an unspecified amount of time will, combined, remove 40% of low quality or unoriginal results. So it’s a 40% improvement compared to when? How are they defining “low quality, unoriginal content,” and would their users agree with the definition?

A net decrease in the amount of spam content on Google can only be a good thing, but this algorithm update also isn’t a silver bullet. Many of the spam issues in Google results started long before AI content became an issue. So far, signs point to this being a general improvement compared to, say, six months ago. It’s not (yet) a return to Google’s golden age, whenever that may have been.

How Can AI be Used Safely for SEO?

Some bells can never be unrung. This is the case for generative AI. We live in the AI era; can it be used safely to create content or otherwise aid our SEO efforts? Or is it better to shelve it altogether when it comes to content?

There is a way to safely use AI to speed up content generation for your website. Rather than using AI as a writer, use it as a writing assistant. For example, here’s my process for writing blog posts with an AI helper:

1. Pre-plan my content. Combine keyword research with industry knowledge and customer research to identify topics that can rank well while providing value to readers. Generate a list of headlines to work from.
2. Turn those headlines into outlines. Create a bullet point outline of each article including potential subheadings and key points to hit on. At this stage, you may start using AI like ChatGPT as a sounding board. Maybe you need one more example for a list in your blog post; ask AI for a suggestion. Perhaps there’s a concept you need to write about that you’re not 100% clear on. Get a summary from AI to help you draft your outline—but be sure to do your own research using trusted sources to confirm! AI doesn’t know anything; while the information provided is generally “ballpark accurate,” AI can hallucinate (read: lie) very convincingly.
3. At this stage I have outlined documents for several blog posts. Time to get to writing! If your outlines are detailed enough, you only need to stitch together each bullet point to create an easy-to-read article. For this part of the process, AI becomes my writer’s block lifeline. Can’t come up with the right word? Paste the relevant bullet point into ChatGPT and ask it to generate some options. You’re probably not going to use its (often incredibly lame) suggestions, but you’ll probably get enough inspiration to get unstuck.
4. When I have a draft completed, I’ll have AI review it. It’s important to know the limits of AI; it’s not going to be a proper human review. It will tell you what you want to hear. Still, asking AI to identify areas of weakness in your article can lead to identifying more topics to include. It’s a nice way to double-check that you haven’t missed anything. As with using AI for basic research, be sure to verify its suggestions against reputable sources.

At no point in this process do I recommend wholesale copying and pasting of AI-generated text. While a single AI sentence, headline, or even paragraph within a human-written article is going to see your SEO penalized, it’s best and safest to commit to human writing with AI assistance.

Conclusion: AI’s frontier is almost settled

The Wild West earned its name fair and square. Chaotic and often lawless, European settlement of the western frontier was a time of unregulated expansion. Like all things good and bad, it came to an end. Hectic frontier towns either vanished or expanded into civilized cities. The long arm of the law took some time to reach as far as the Pacific, but reach it did.

Today, the AI frontier is in the midst of its settlement. After a lawless first year, AI tools face more scrutiny from users, businesses, and governments alike. The awe that tools like ChatGPT were met with upon initial release is settling down. They say that in the Gold Rush, the people who sold picks and shovels were the only ones to make any money. As some early adopters of AI content creation are finding out, life on the frontier can prove perilous.

Over the coming months and years, the online world will find its equilibrium. The role of AI in our day-to-day lives is a certainty, but the extent of that role is still up for debate. For marketers, AI may not be the cure-all many hoped it would be in 2022.