The cost of producing a 1,500-word article has collapsed to somewhere near zero. That’s the supply shock, one my friend Mark Schaefer has talked about for years. The more interesting question today — the one most marketing leaders haven’t priced correctly yet — is what that collapse does to everything else.
When a factor of production goes free and infinite, value doesn’t disappear. It shifts. And in marketing today, it’s shifted somewhere most content strategies aren’t looking.
Academic evidence has seen this coming. A National Bureau of Economic Research paper using Pixiv data shows that generative AI is crowding out human creators. Ahrefs data shows that 86.5% of top-ranking pages now contain some amount of AI-generated content. And Graphite.io found that the total quantity of AI-generated articles probably surpassed the quantity of human-written articles published on the web within the last couple of of years.
Additionally, research published in Nature on "model collapse" — the degraded outputs that occur when AI trains on AI output — singal a related, and more problematic reality for marketers: a "collapse to uniformity.” That’s the steady, unrelenting increase in “textual similarity” across the web since AI started publishing in the 2010s. Those similar, AI-generated outputs have accelerated ever since ChatGPT’s earliest models, and are projected to reach 90% saturation around 2035.
If that forecast holds, we’re adding roughly 10 points of uniformity — bland, boring, blah content — every year. And that means the window for you to differentiate is not some theory. It’s real. And it’s closing on you right now.
The result is two things happening simultaneously:
- AI-generated content is increasingly good enough that it saturates every channel.
- And human beings — who are, as I say, "pretty good bullshit detectors" — are beginning to flag content they don’t like as "AI slop.”
That second reality isn’t happening just when content was generated by a machine. It’s also happening when it doesn’t sound human enough, when it’s too corporate, too polished… too fake.
It’s fairly likely that at least 25% to 30% of your customers will actively demand authenticity within the next two to three years… if they aren’t demeaning it already. My two-to-three year forecast isn’t pulled out of thin air — it’s the sheer math of 10 points of AI detection per year building off a base of roughly 10% today.
This episode of Digital Reset with Tim Peter identifies the three specific assets AI cannot reproduce:
- Proprietary data
- Original examples
- Your expert voice
We also share a three-question framework for auditing whether your content strategy is actually building any of them.
The closing argument is my own practice of writing our podcast scripts by hand, despite having AI tools running in the background: not because I can write faster than a machine. We all know that’s not true.
Instead, it’s my experience, my beliefs, my humanity are the scarcest resource our content can put to work. That the distinction that matters — separating what is abundant from what is actually scarce. It’s the core claim I’m making this time. And it’s the one that will determine which brands still get seen in three years time… and which will have blended into the background.
Key Insights for Marketing and Business Leaders Navigating AI Content in 2026
In this episode, we break down:
- AI didn’t kill content marketing — it repriced it. The cheap parts (generic explainers, commodity how-tos, undifferentiated articles) are now worth close to nothing, because anyone using AI can produce them in seconds. The expensive parts — your proprietary data, lived experiences, and genuine expert voice — have become more valuable, not less. Using AI won’t make you fail. But spending your content budget on the wrong side of that line will.
- "AI slop" is the uncanny valley of content — and your customers’ bullshit detectors are already activating. People flag content as AI-generated not just when it actually is, but when it fails to sound human enough. As customers get increasingly sensitive to “AI slop” — probably by 10 points or so per year — the brands relying too heavily on overly templated, indistinct, and impersonal content will find themselves on the wrong side of a widening credibility gap. Somewhere between 25 to 30% of customers will demand clear authenticity in the next two to three years… if they’re not already.
- "Collapse to uniformity" is the structural threat underneath "AI slop." The Nature paper on model collapse gets most of the attention, but the follow-on research on “textual similarity” is the more immediately relevant fact. Content on the web keeps getting steadily more similar, and has only gotten worse ChatGPT emerged on the scene.. Researchers project 90% saturation by 2035 — roughly nine years away. That means the differentiation window is not just some theoretical abstract. It is measurable, and it is closing.
- The three assets AI cannot fake are proprietary data, original examples, and expert voice. Data that only you have about your customers, your market, and your industry is yours alone to report. Original examples from real customers carry the credibility of lived experience that no AI can generate. And an expert voice means being willing to make a specific, named predictions and opinions — ones you’re willing to be wrong about — because taking that risk is exactly what makes you worth listening to. Generic best practices and how-to content is dead. Your truth, your actual opinions, that you’re willing to own, is rare.
- Three questions will tell you where your content strategy actually stands. What is your ratio of proprietary to commodity content in the last 90 days? When a prospect asks ChatGPT, Gemini, Claude, or Perplexity about your category, what specifically about your brand shows up — and is it something only you could have said? Who on your team is your named voice, and are you amplifying their signal or burying it it in generic content calendar outputs? The answers to those three questions are the most useful content audit you can conduct.
- The answer is not to publish less — it’s to publish more of what’s scarce. This episode is not about cutting volume. It’s about redirecting where you put your efforts. Publishing more of the content only you can produce — and less of the content that any AI could produce — is the allocation of time and resources that matters most. The brands that figure this out in the next two to three years are the ones customers will ask for by name. The ones that don’t will blend into the background.
Whether you’re a CMO deciding where to concentrate your content budget in 2026, a marketing leader who keeps being asked about AI, or a brand that’s already noticed its content working less well than it did a year ago, this episode gives you the framework to understand why… and what you can do about it.
AI Made Content Free. Here’s What It Made Priceless. (Digital Reset Episode 492) — Headlines and Show Notes
Show Notes and Links
