Around the time ChatGPT launched, a lot of marketers saw the same opportunity. Suddenly, we had tools that could write passable blog posts in minutes (they were a bit slower back then). For capacity constrained content teams, it seemed like an obvious way to skyrocket production. More posts with more keywords could mean climbing the keyword ladder. And with executive teams constantly putting pressure on marketing teams for more, if you could generate hundreds of articles a month, why wouldn't you?
Some of those marketers quickly woke up from their fever dream and realized when something seems too good to be true, it usually is. Still, many pursued these types of strategies and some continue today. We even looked into the strategy ourselves in the early days. We tried a formula of taking an idea, feeding it research compiled by AI tools, and letting it write something that felt like it had a unique spin. Sure, it read like a listicle, but it had all the patterns of high performing content that could get clicks. And a lot of smart people like that.
But then reality came crashing down. First, it became clear that, just like every past low-value short-term growth strategy, that Google planned to crack down on AI slop articles. Second, the AI tools themselves got better at finding relevant information and people got good at using them to find what they were looking for. And those tools, like Google, value delivering good quality content to users. The result is that AI slop content didn't seem like such a good idea for a variety of reasons.
Where we've landed now is an understanding that the underlying principles of good content didn't change in the age of AI. They became more valuable.
The double failure of content without perspective
People visit your website for a reason. They have a question, a problem, or they're trying to figure something out. You get one shot to show them you understand their situation and that you're worth their time.
Content that was generated to fill a keyword gap doesn't do that. A visitor who lands on generic, obviously AI-written text doesn't think "well, at least they're publishing a ton of stuff." They think your company is as detached and generic as the content you put in front of them. Then they leave.
That's the first failure, it's immediate and it hurts. The second won't be seen immediately but it's going to hurt a lot more.
Search engines and AI tools don't want to deliver bad content. Google has spent the last four years tightening their approach to this. Their Helpful Content system got integrated directly into core ranking in March 2024. The March 2025 core update explicitly rewarded depth, originality, and firsthand experience. By mid-2025, Google was penalizing websites for what they call "scaled content abuse," completely deindexing sites that were mass-producing AI articles. Their quality raters now specifically assess whether content is AI-generated, and rate it lowest when it lacks effort and originality.
The pattern has never reversed. Every update moves in the same direction, reward content that's worth someone's time, filter out the bad and lazy stuff.
Ahrefs studied 300,000 keywords and found that AI Overviews correlate with a 58% drop in clicks to the top-ranking page. Eight months earlier, that number was 34.5%. It's accelerating. Seer Interactive's research across 42 organizations tells the other side of the story, that brands that get cited as sources in AI Overviews earn 35% more organic clicks, while non-cited brands take a big hit. The dividing line is whether your content is source material people actually want or low quality filler used to get clicks.
So the volume strategy fails twice. It doesn't convert the visitors it gets. And over time, it stops getting them at all.
What "AI slop" actually means
I had more than three clients send me links to services promising a hundred SEO articles a month. The rep who contacted them claimed the articles would be fully brand-aligned, completely accurate, and unique. After looking at a few of these, I advised our clients to avoid them. One pushed back, so I challenged them to sit down and really read one of those articles and tell me what they found valuable.
They pushed back that it wasn't about the quality of content, but about getting people to click through to their website. This is bad strategy.
If you're more concerned with beating people over the head with the fact that you exist than providing them value, you're taking your brand down a bad road with or without AI.
At the same time, there's nuance that gets lost in the "AI slop" talking points. Some people use that phrase to mean anything AI-generated. But AI doesn't own the rights to creating slop content, humans have been doing it for years. Those listicles I mentioned the AI content sounding like were around long before we had ChatGPT. We taught the AIs how to make this junk. The tools just make it much, much faster.
The problem, then, isn't AI or not AI. The problem is publishing content that doesn't have a real perspective behind it. Content that doesn't answer a question, solve a problem, or give the reader something that they came for.
This pattern always repeats. Right now, there's some AI shortcut that seems appealing in the moment but will look just like this in a year or two. The people rushing to exploit the latest capability without asking if the content actually helps anyone are making the same bet the hundred-articles-a-month crowd made in 2024. The technology changes, but the mistakes rhyme.
AI as accelerator, not author
This doesn't mean AI has no role in content. It means the role is different than many companies think.
Every company has domain experts. Engineers, consultants, account managers, people who work on the product or with customers every day. They have perspectives that would make genuinely useful content. But they're not in the marketing department, and the marketing department would love to get those viewpoints out into the world.
In the past, we've solved this problem by teaching our clients how to run expert interviews (or running them ourselves). A marketing person prepares a script, interviews the expert, and creates content from the conversation. There are two problems with this tactic. The marketing person isn't the domain expert, so their questions often don't pull the most valuable insights. Then they have to write the piece without the underlying expertise, so the best points can get lost in translation.
We built an AI interview agent that addresses both problems. The agent is trained on the company's existing content and technical documentation, so it can prepare questions that meet the domain expert where they are. The domain experts get to nerd out with an AI interviewer, digging deep into the value they deliver customers. The tool then captures the unique perspective required for truly great expert content. Then the drafting process weaves the expert's perspective with the company's brand voice and supporting documentation. The output is a draft that's often tied directly to real customer concerns or solutions the company has built.
Even though this is already often a better end product, it's not ready to ship. It still needs a human-in-the-loop.
AI writing has a tendency to be overly polished. It avoids repeating words in a way that no human actually does. It follows patterns that are technically correct and also boring. There's a smell to it. Readers pick up on it, consciously or not, and it feels inauthentic. Good writing is flawed, that's what makes it human.
The best way I've found to demonstrate this is to ask AI to write an article about a topic you're interested in (can be work related or not, fiction is even fine). Something you'd really enjoy reading. Then, sit down and try to read it like you'd read a real book. I would bet that you won't make it too far before you're ready to put it down. It's a topic you care about, written for you, and it's boring as hell. The mechanics are there, but the life has been squeezed out by perfection.
I use AI to support my writing process and I've found an editing approach that works well for me. I read what the AI drafted, paragraph by paragraph, then I say it back the way I'd actually explain it to another person. Sometimes out loud using voice-to-text, sometimes rewriting by hand (artisanal blog writing, some might say). That's where my voice comes back in. The ideas were captured by the expert interview, and that's great, but our brand and my unique writing style were lost. I've fine-tuned models on hundreds of pieces of my writing and the AI just can't get there. I have to edit heavily to get it where I feel it needs to be.
That's where the brand shows up. Every organization will have their own version of this, their own style and their own perspective. The point is that the human pass is required. It's the whole reason the content is worth reading.
The endgame
The endgame is the chatbots. When someone asks ChatGPT a question about a problem in your domain, you want to be the answer it pulls from. Even if the tool answers the question directly, the user sees your brand attached to the source and that builds trust. You're the authority, even the AI says so.
You don't get there by publishing content that any of these tools could generate without ever mentioning your name. You get there by answering real problems that very few other people have the perspective to answer. Content that's specific to your expertise, tied to the work you actually do, written in a way that could only come from your organization.
It's our AI interview tool and the requirement of human editing. Those are the things no one else is going to be talking about when they have AI write a slop article on AI slop articles.
We rebuilt our own website recently with AI support across writing, development, and design. Always with a human in the loop. Always a person making the decisions. The early analytics show higher qualified leads and stronger lead quality and total value. AI accelerated every aspect of the work (cutting the build timeline in half) and helped improve the quality. But the decisions, the perspective, and the final judgment were always owned by our team.
Higher quality at a faster pace is a real advantage everyone should be taking advantage of. But more quantity with no quality is an expensive way to damage your brand.
Write for people. The algorithms will catch up.