The Search Singularity: How to Win in the Era of Infinite Content


Weird things happen in the heart of a black hole.

Huge stars collapse under the strength of their own gravity to become one-dimensional points. Spacetime reaches infinite curvature. Matter is crushed into infinite density. The laws of physics that govern the rest of the universe grind to an unimaginable halt.

The very center of a black hole, the precise point where all of this funkiness happens, is called a gravitational singularity. I bring it up because there’s a similar (if less cataclysmic) thing happening in content marketing. You can think of it as the search singularity, and it’s the moment when we reach a critical mass of copycat content—when the search results become so swamped by samey blog posts and duplicative guides that the regular rules of content marketing break down.

While gravitational singularities can be traced back to dying hypergiants, the search singularity has its roots in one of the most exciting technologies in human history: AI.

The Search Singularity

In May 2020, we spent time tinkering with OpenAI’s GPT-3, an AI model focused on writing. It was uncannily good—able to generate flowing, readable, human-like prose (even poetry) from a handful of input prompts. We spent a few meager hours toying with it, and we were still able to bootstrap an article that read just like a human writer (albeit not a very good one) had created it.

In the months that followed, we saw a wave of new tools released that use the model’s API to create headlines, product copy, social copy, even articles entire from a few sentences of input material. They serve as the counterpart to a suite of existing tools that use older AI models, like IBM’s Watson, to help structure and search-optimize content.

For many people, this next iteration of AI-enabled writing represents a potential process improvement, a way to speed up the content marketing workflow and get to a finished draft faster. For others, it’s an overhyped gimmick with limited real-world application, another marketing buzzword in the vein of Clubhouse and VR.

The truth is neither. AI writing tools mean that the marginal cost of a blog post is nosediving from multiple skilled person-hours to minutes spent in a freemium SaaS product. In the near future, any company with a modest budget and a functioning internet connection will be able to pump out truly mind-numbing volumes of content.

Some of these articles will be pretty good. There will be people that use AI tools as a jumping-off point, using it the way a skilled craftswoman uses any tool—to turn something good into something great. Copy.ai can function as a powerful brainstorming counterpart. Writer help seamlessly implement your company style guide, even functioning as a tone-of-voice editor. Other tools can flag biased language and improve your grammar.

But there will also be people that use these tools wholesale, without any care, attention, or thought. They’ll be able to plug in a topic or target keyword, and the AI will spit back an article in mere moments. They will drown the web in huge volumes of garbage.

This is the difference between AI-assisted writing and AI-generated writing. The former will lift content marketing to new heights; the latter will trigger the “search singularity,” an implosion of homogenous, copycat content on an unparalleled scale:

  • Passable articles will be created without skilled human input, research, and process, allowing companies to scale content to an astronomical magnitude.
  • Every article will see competition from hundreds, if not thousands, of other articles, and uncontested search results will become a thing of the past.
  • Off-page ranking factors will become disproportionately important in response to every article containing the same remixed information as every other.

In the next few years, our single biggest task as content marketers will be to rise above this tide of mediocrity.

Past the Event Horizon

While it may not feel like we’ve reached this point, we’re already on an unswerving trajectory. To use more physics terminology, we have passed the event horizon, the blackhole’s inescapable boundary from which not even light can escape.

AI is already being used to brainstorm, workshop, and refine content. It’s a slippery slope straight towards full-bore AI-generated content marketing:

  • AI is already helping to structure and search-optimize articles (Clearscope)
  • AI is already helping to identify gaps in subject matter knowledge (MarketMuse)
  • AI is already being used to flesh out bullet point outlines and get to the first draft faster (Copy.ai)

To quote William Gibson, “The future is already here—it’s just not evenly distributed.” It’s only a matter of time before the adoption of AI writing tools reaches a critical mass, and we begin to feel the effect in the form of vastly oversaturated search results.

Don’t believe me? Try these critiques on for size.

“But it won’t be good enough to rank.”

The elephant in the room here is quality: it’s easy to argue that AI-written content isn’t good, and certainly not good enough to rank; that it simply dumps connected ideas together on a page and uses passable sounding phrases to connect them together. It creates a simulacrum of good writing, something that looks good at first blush but falls apart on closer inspection.

I agree—but this is also how the vast majority of the web’s SEO content is currently written. Read through the top ten search results for virtually any keyword, and the chances are good that these “top-performing” articles are rife with the same problems that plague AI-generated content. No narrative. Repetitive information. Unoriginal formats.

Factors like on-page SEO, domain authority, and topical relevance already play a disproportionate role in dictating search rankings. Bad content is already perfectly capable of ranking. It does every day. AI-written content will follow suit.

Even if you don’t buy this argument, it’s important to avoid underestimating the rate of progress in AI. Even if AI-written content still requires a huge amount of human input to become functional, every refinement to the model, every additional data source, and every additional guardrail added to the frontend means that the amount of input required is falling day by day. If it isn’t good enough to rank now, you better believe it will get good enough in short order.

“It’ll get penalized for plagiarism.”

It’s also important to avoid conflating AI-written content with older types of “spun” content: lightly reworked versions of existing content. Spun content often falls foul of plagiarism detectors, and Google is good at detecting many forms of spun content (it’s been working at it since 2012).

But GPT-3 does not spin or plagiarize—it uses the vast corpus of literature at its disposal, coupled with your input prompts, to create a sequence of words that fit with the model’s understanding of the established rules of writing. Much of what GPT-3 generates has never been written before. It is by most measures original content.

“You still need strategy.”

When we think about search content, there’s obviously another marginal cost to consider: strategy. Conducting keyword research is a laborious, skilled process, requiring prioritization, competitor analysis, intent categorization, and judgments around feasibility.

But to cut the Gordian knot, why bother with strategy when you can create hundreds of blog posts at a time? Why bother with prioritization when you can target every keyword? Why bother with competitor analysis when you can just copy your competitors in a fraction of the time?

This is a facetious argument with a serious point. Most companies will continue to understand the importance of strategy, and they will create the best content with the greatest return on investment—but there will be a subset that chooses to use content marketing like a fire hydrant, using AI to target keywords indiscriminately.

Even if this is only a small proportion of content creators, AI will allow them to be so prolific as to cause a real headache for the rest of us. Search results will grow more cluttered. Readers will grow more skeptical. There will be ever more noise and ever less signal.

“Google will adapt.”

SEO has always functioned as a game of call-and-response. It’s likely that Google will find new ways to exclude weak AI-written content from the SERP by devaluing traditional on-page signals and relying more heavily on signals like information gain (more on that below) and authority. But this will take time—time in which your competitors could be gaining a stranglehold on the search results.

There is no “undoing” this shift, and the sooner we can embrace this mindset, the better. We just have to adapt and learn how to win in the era of infinite content.

How to Win in the Era of Infinite Content

If you open yourself to the idea that we’re on the cusp of a staggering uptick in AI-fueled content creation, most of which will be functional but unremarkable, one clear realization emerges: Quality is a powerful differentiator today, but it’s about to become vastly more important.

One by one, information asymmetries and operational moats have been eroded. It’s harder than ever to be first or faster, but the incentive to be better is the greatest it’s ever been. Your job is to create a vast gulf of quality between AI-generated content and your content.

Imagine you now live in a world in which most search results are contested by 100 procedurally generated articles. What would you need to do in response? Here are three things you can start doing today:

1. Focus on “information gain” in every article you create

The Achille’s heel of any AI model is (for now) research. By definition, any article created by GPT-3 is pulling from the same vast database of information as every other article it creates. It can remix and rewrite, but it cannot research. It can’t interview people. It can’t conduct surveys. It can’t easily evaluate existing content for information gaps.

One sure-fire way to maintain an edge over AI-generated search content is to write with information gain in mind. Information gain is a measurement of the new information provided by a given article, over and above the information present in other articles on the same topic. Information gain scores can be used by Google to influence search rankings: an article that adds something new to the discussion may rank higher than an article that repeats the same information as others.

Primary research is the ultimate form of information gain. By adding original survey data, expert quotes, or addressing neglected parts of the topic, your content is adding brand new information to the discussion—something which AI content can’t yet replicate.

2. Diversify beyond search and invest in thought leadership

We’ve focused this article on search content because it’s here that AI will breach first, for the simple reason that today’s search content is not dissimilar to AI-generated content. Tactical content (like how-to’s and standard processes) are easy for AI to reproduce—but things like opinion and narrative are harder.

That’s where thought leadership content comes in. Thought leadership generates awareness and builds trust by sharing a company’s “earned secrets:” the unique experiences, insights, and perspectives that only that company has. This can take many forms, each of which is next to impossible to create using AI alone:

  • Counter-narrative opinions (exploring why the status quo is wrong or flawed)
  • Personal narrative (sharing your own lived experiences)
  • Network connections (sharing the experiences of people within your network)
  • Industry analysis (using our knowledge to add context to industry events)
  • Data storytelling (revealing the insights hidden within data—yours, or publicly available data)

Dedicating a small portion of your budget and mental energy to other forms of content marketing is always a good thing. Like the concept of hedging in finance, the strengths of one type of content (like thought leadership) can offset the weaknesses of another (like search content), creating a diversified, lower-risk “portfolio” of content that’s less susceptible to changes in search algorithms. Thanks to AI, the model is less “library versus publication,” more “library and publication.”

3. Share the same information but create a new experience

Even if we assume a bleak dystopian future where every article shares precisely the same information as every other, there are still opportunities to create a different experience. Take Shakespeare’s Hamlet by way of example: it’s a single, classic story that has been told in hundreds of different ways, from the 15-minute Hamlet to the Klingon Hamlet to the Puppet Players’ Hamlet.

The core information—the story—is the same, but the experience is vastly different. Different themes are emphasized in each portrayal. Minor variations in acting and speech create new dynamics between characters. The tone varies wildly from melancholy to comedic to absurd. Even if the audience already knows the story of Hamlet, they can still take something new and interesting away from each viewing.

Content marketing has similar variables at its disposal. The same information can be shared in a way that’s either punchy and concise or eloquent and laden with story. You can focus on metaphor or real-world example. You can highlight the opinion and personality of a single person or strive for an ultra-authoritative Wiki-style tone of voice.

Brand and personality are powerful forces. Even if you’re forced to compete with a dozen other articles, each sharing the same information, you can still create a unique experience that will appeal to your target audience.

Futureproof Your Content

The web is already awash with duplicative, unoriginal content (we call it copycat content). This was problematic even when content marketing was in its infancy, but thanks to AI, it’s becoming ever easier to churn out vast volumes of mediocre content. In short order, this content will be able to rank, and there’s a very real risk of most search results becoming a battleground between AI-generated content—a search singularity.

Even if you don’t buy the argument that we’re on the cusp of a search singularity, it’s worth recognizing that these problems already exist today. The issues we’ve outlined here are simply an acceleration of an already existing trend; that of ever more noise and ever less signal.

AI or not, these tactics—focusing on “information gain,” experimenting with thought leadership, and creating new content experiences—are growing more important for standing out from your competitors, carving out a lucrative niche in more crowded search results, and futureproofing your content marketing.