The hidden cost of AI-generated articles at scale

The hidden cost of AI-generated articles at scale is this: the failure doesn't show up when you publish, it shows up months later as a mess you pay someone to untangle. The upfront cost is almost nothing - that's the trap. The real cost is deferred, and it's usually bigger than what you'd have spent doing it properly the first time.

I want to be precise here, because 'AI content is bad' is lazy and wrong. AI content isn't inherently bad. AI content at scale, published without human judgement, creates specific structural problems that no tool reliably prevents. Let me name them.

Problem one: cannibalisation

Cannibalisation is when two or more of your own pages compete for the same keyword. Google can't tell which one you want to rank, so it hedges and half-ranks both - and 'half of two' is worse than 'all of one'. When you generate articles at volume, you produce near-duplicate angles on the same topic constantly, because the tool has no reliable memory of what it already covered or judgement about whether two pages truly serve different searches.

A human writer would notice they'd already covered 'commercial cleaning in Auckland' and either consolidate or deliberately differentiate. A generation engine running at scale just makes the fourth version. This is a big part of why publishing more pages can make your rankings worse, which still surprises people who assume more content always helps.

You don't get billed when you publish. You get billed when someone has to work out which pages to delete.

Problem two: thin content and the indexing tax

The second problem is subtler. Search engines have a rough budget for how much of your site they'll crawl and index, and a growing sense of your site's overall quality. Flood it with a hundred pages that each say a little about a lot and nothing decisive about anything, and two things happen. Some pages never get indexed at all - the engine looks, shrugs, and moves on. And the pages that do get indexed drag your average quality down, which can affect how the rest of your site is treated.

So you've spent effort producing pages that either don't exist as far as Google's concerned, or actively make your good pages look worse by association. That's not a neutral outcome. That's negative return on the work.

Why no tool reliably prevents this

The tools will tell you they handle this - deduplication, quality filters, internal-linking logic. And they help, a bit. But the underlying judgements - is this page genuinely distinct, does it deserve to exist, which existing page might it undercut - are exactly the calls I don't trust a tool to make at mass scale without oversight. The whole premise of scale is removing the human. The whole solution to these problems is the human.

That's the contradiction at the heart of it, and it's why I'm sceptical of the autonomous 'set and forget' tools that promise volume without the cleanup. You can have scale, or you can have reliable quality control, but the moment a tool promises both with no human in the middle, someone is going to be paying to untangle the difference.

The untangling bill

So what does the deferred cost actually look like when it lands? Usually some combination of:

• An audit to work out which of your pages are competing with each other.
• Decisions on each one - delete, merge, redirect, or rewrite - which is slow, careful work.
• Rebuilding internal links that pointed at pages you've now removed or merged.
• A recovery period while the engine re-evaluates a site that suddenly looks very different.
• The opportunity cost of all the months you spent 'producing content' that netted out negative.

Add that up and the cheap option wasn't cheap. It was a loan with the interest hidden in the small print.

What I'd do instead

Fewer pages, each one deliberately chosen and clearly the best answer to a real search, with internal links placed on purpose. That's slower to produce and far cheaper to own, because there's nothing to untangle. AI absolutely helps me produce those pages faster - it does the drafting and formatting - but a human decides which pages exist and how they connect.

And for a lot of local businesses, the honest answer is that they don't need a content programme at all, at scale or otherwise. Before you automate an article factory, it's worth asking whether blogging is even worth it for your kind of business. Producing the wrong thing efficiently is still producing the wrong thing - you've just made the eventual cleanup bigger.

I'd be happy to be proven wrong as the tools mature, and I don't know what I don't know. But right now, 'cheap AI articles at scale' reads to me like a cost you can't see yet, not a cost you've avoided.

If you decide a content programme genuinely is right for your business, the four structural gaps that create these hidden costs in the first place are worked through in order in the honest guide to using AI for your website content.