I Said AI Wouldn’t Kill SEO. Data Finally Agrees

“Video killed the radio star” has got nothing on SEO. Voice search was supposed to kill SEO. Featured snippets was supposed to kill it. Now AI has killed it, apparently for the third or fourth time this year.

SEO has died more times than a soap opera villain. I’ve watched this rerun for 20 years. Just weeks after ChatGPT launched, I put my position in writing while the panic was at its loudest: AI won’t kill SEO, and mass-produced AI content will get punished.

Three years later, the receipts are in. Let’s walk through them.

The Paper Trail Starts in January 2023

Two months after ChatGPT launched, everyone lost their minds. Marketers declared SEO dead. Copywriters updated their resumes. Agencies pivoted their entire pitch decks overnight.

On January 10, 2023, I published a post titled “ChatGPT is going to kill SEO.” Spoiler: the body of that post said no, it’s not, and called out the bandwagoners directly. My reasoning was boring and practical. Someone still has to write the prompts. Someone still has to revise the output. Someone still has to fact-check the claims. And a tool that shuffles existing content into new arrangements doesn’t birth original ideas.

That June, I followed up with “Will AI kill SEO? No.” Same answer, more detail. I’d tested ChatGPT and every SEO content tool before it. Jasper had been “the future” less than a year earlier.

Content spinners were “the future” a decade before that. The tools change. The pattern doesn’t.

Then, in August 2023, I made the prediction that got me the most pushback, “Google is going to punish AI content.”

To prove a point in how easy it was to detect patterns in AI-produced content, I’d surveyed a group of “experts” on a topic, and half of them sent back nearly identical answers. Different experts, different “special prompts,” same regurgitated output.

If a regular guy like me could spot the duplication with his own two eyes, Google’s systems are infinitely smarter and would have no trouble at all.

The comments section disagreed. Loudly. I was told detection was impossible. I was told I didn’t understand the technology.  I was told prompting had evolved, and you just needed that super duper, special secret prompt.

Okay.

Then Google Made It Official

On March 5, 2024, Google announced a core update and new spam policies built to reduce low-quality, unoriginal content in search results by 40%.¹ Their spam policy update specifically targeted “scaled content abuse,” which is the polite term for pumping out mass-produced content to manipulate rankings, no matter what tool produced it.²

Sites built on unmoderated AI output got crushed. Some got deindexed entirely. The businesses that had fired their writers to “save money with AI” suddenly needed those writers back, plus a recovery specialist, and a prayer.

I’d seen this exact algorithm roll out before. In 2011, Google’s Panda update hit a documented 11.8% of all searches and wiped out websites built on thin, spun, mass-produced content. The corner cutters got flattened. The people doing original work barely noticed. My business grew through Panda because panicked owners needed someone to dig them out of the hole their previous SEO buried them in.

Swap “content spinning” for “AI generation,” and 2024 was a remake of 2011, shot for shot. Mass-produced content at scale is nothing new. Only the software gets a fresh coat of paint.

Human-Generated Content Performs Better Than AI

Predictions are cheap. Data is better. So here’s the data.

Neil Patel’s agency, NP Digital, ran one of the largest head-to-head tests published on this question. They created 744 articles across 68 websites, splitting the work between human writers and ChatGPT, then tracked traffic for five months.³

Yes, the AI was faster. It produced an article in roughly 16 minutes, compared with 69 minutes for a human writer.³ If speed were the goal, AI wins by a landslide.

Speed was never the goal. At least, to a qualified SEO it should never have been.

Traffic, trust, and revenue are the goals. By month five, the human-written articles were pulling 5.44 times more traffic than the AI-written ones. The human content grew steadily month over month, while the AI content bounced around and stalled.³

Here’s the part that should end the debate for anyone running a business. Even after accounting for how much longer humans take, human content still won on efficiency: 4.10 visitors earned per minute of writing time versus 3.25 for AI.³ The “slow” option produced more results per minute invested.

Cheap content that nobody reads isn’t cheap. It’s just invisible.

AI Content Is the Floor, Not the Ceiling

Rand Fishkin, who built one of the most successful SEO software companies before becoming a full-time data analyst, framed it well back in April 2023. AI-generated content is the new floor. Anyone can make it, most of it is passable, and that’s exactly the problem. When everyone has access to the same “pretty good,” pretty good becomes worthless.⁴

Fishkin pointed to three things humans still do that language models can’t: genuine emotion, actual novelty, and creative insight.⁴ A model trained on existing content can only remix existing content. It has read millions of articles about cookies. It has never walked into Grandma’s house during the holidays and smelled them baking.

That’s not a cute detail. That’s the whole game. I have a client who sells home-baked goods, and no AI on earth knows their founder’s origin story, their weird oven quirks, or the customer who orders the same box every Christmas for her deployed son. Those stories are the one asset your competitors can’t copy and a language model can’t fabricate, because it doesn’t know them.

The same logic explains why the doom predictions about search itself flopped. Google search volume grew 21.64% the year after ChatGPT supposedly replaced it, and Google still handles hundreds of times more searches than AI chatbots.⁵ Bing has more search volume than ChatGPT and Perplexity combined. Bing…

Of the small slice of queries AI tools do handle, only about 30% are transactional. Most people are asking AI to “write me a thing,” not to find a business to buy from. Buyers are still on Google. The commercial action never left.

The AI Bros Are Speedrunning Old SEO Mistakes

Watch what’s happening inside AI search right now, and you’ll see the sequel already filming.

Lazy optimizers flooded Reddit to influence AI answers, so that door closed.

Then they spammed syndicated press releases, and now wire-service content barely shows up in AI citations at scale.

In early 2026, ChatGPT hid its query fan-out data, the exact metadata corner cutters were using to reverse engineer sources. Two hacks, gone in a week.

Meanwhile, the content getting cited by AI tools is original editorial from real brands with real expertise. AI systems are drawing harder lines between content they trust and content they see as filler, the same way Google spent 20 years learning to do. Every shortcut crowd eventually rediscovers the same lesson: platforms reward what their users actually value, and their users value original, useful, human work.

Here’s my favorite piece of irony from this whole saga. I launched a sister agency focused on AI optimization, and our first formal AI optimization lead came from a Google search. The way you show up in AI answers is a strong technical foundation, genuinely useful content, and credibility signals from other trusted sources. That’s a description of SEO. It always was.

Your Voice Is the Moat

I’m not anti-AI. I use it, my team tests every new tool that comes out, and it’s genuinely useful for research, outlines, data analysis, and grunt work. Even Patel’s study concluded the same thing: use AI for brainstorming and heavy lifting, then let humans do the writing.³

What I’m saying is what I said back in January 2023, when saying it was unpopular. Since ChatGPT launched, my team has published well over a thousand posts and have used AI to fully draft exactly zero of them. In that same window, leads from my content roughly doubled without me increasing volume.

What I’m highlighting is too many people are overlooking the opportunities outside AI.  The opportunities in your own unique stories. Authentic, human quality beat AI quantity in my own numbers before any study confirmed it.

The takeaway for you is bigger than one prediction aging well. Every wave of automation creates the same opportunity: while the majority races to the bottom, publishing identical, mass-produced AI slop, the minority that keeps telling original stories in a recognizable voice stands out more than ever.

AI lowered the barrier to entry. It also raised the value of everything above the floor.

So don’t fall into the AI sameness trap. Write the story only you can tell. Share the numbers only you have. Say the thing your competitors are too generic to say.

The machines can’t take your voice.  Only you can forfeit it.