Will AI Replace Bloggers?

The number that stops most bloggers is not a headline about AI. It is a line on their own analytics.

Impressions up. Clicks down. The post is being shown to more people than ever and visited by fewer of them, month after month, and nothing about the post has changed.

You know what that means before anyone explains it to you. Somebody is getting your answer without arriving at your site.

The clicks went before the writing did

The Pew Research Center followed almost 69,000 Google searches from 900 American adults in 2025. Where an AI summary appeared in the results, people went on to click a regular search link on 8% of visits. Where no summary appeared, 15%. Roughly half the onward traffic, gone in the space of a redesign.

That is what has actually happened to blogging. Not a machine that writes better than you, which was never the threat anybody should have been worried about. A machine that answers the question above your link.

Meanwhile the writing profession itself has not fallen off a cliff. The Bureau of Labor Statistics projects writers and authors, the nearest occupation it tracks, to grow about 4% through 2034. Blogging is not in collapse. Its distribution is.

Something we keep running into when we look at blogs in this position is that the drop is never even across an archive. The quick explainers and the definition posts fall off a shelf. The write-up of the thing you spent a weekend actually doing barely moves.

What AI cannot do on a blog

Three things, and prose quality is not one of them.

The first is doing the thing and reporting back. A model can produce a fluent paragraph about a stroller it has never pushed up a curb, and that is precisely why every one of those paragraphs reads identically. What it cannot produce is the fourth week, when the wheel started clicking.

The second is being the source an answer gets built from. Every summary is assembled out of something somebody published, and original testing, measurements, and photographs are what put a site in that pool rather than downstream of it.

The third is being a person readers come back to. Nobody subscribes to a summary, forwards one to a friend, or emails one to ask what they should buy instead. That relationship took you years and it is the asset the search box was never in charge of.

The response that makes it worse

Here is where a lot of bloggers go wrong, and it is worth naming because the instinct is so reasonable.

Traffic falls, so you publish more. AI makes publishing more almost free, so you publish a lot more. Twelve posts a month instead of four, all competent, all assembled from what was already online.

You have just produced more of the exact thing that got replaced.

What surprises us most is how rarely anyone frames it that way. Volume stopped being scarce the moment a decent draft cost four minutes. What is scarce now is being the only site that knows something, and no amount of publishing speed gets you there.

If your own archive is telling you a version of this story and you can’t quite read it, the 3-minute readiness check walks through what your posts are made of rather than how many of them there are.

Be the thing the answer has to cite

Google’s own guidance on helpful content asks publishers to self-assess on whether a page shows first-hand expertise and offers original information rather than a rewrite of other sources. Its position on AI-written content is that how a page was produced matters less than whether it is useful. Read those two together and the incentive is not subtle.

Nobody outside the company can promise you a ranking. What they can tell you is which direction the door is open.

So the work changes shape. You spend fewer hours writing and more hours generating things worth writing about. You test the product for six weeks instead of comparing specification tables. You photograph your own failures. You publish the number you measured, the batch you threw out, the thing you believed two years ago and no longer do.

Then you use AI for everything around that, which is most of the evening anyway. Start with a question about your own archive.

Here are three posts of mine that held their traffic this year and three that lost the most. Tell me what the ones that held up contain that the ones that dropped don’t, looking specifically at whether each post has anything in it that could only have been written by someone who did the thing. Be blunt.

The answer is usually uncomfortable and takes about ninety seconds to arrive.

If you want the whole workflow rather than one prompt, the AI for Bloggers course runs from topic research through drafting in your own voice to refreshing an archive you already own, which is where most of the recoverable traffic is sitting.

So will AI replace bloggers?

It will replace the interchangeable post. That was already happening before any of this, and the summary just finished the job faster than expected.

It will not replace the person who owns a kitchen, a workshop, a trail, a spreadsheet of things that went wrong, and eleven years of answering the same reader question in a slightly better way each time.

The uncomfortable part is that those two kinds of blogging have been sitting inside the same archive, on the same domain, for most of us. One half is being taken and the other half is quietly becoming more valuable, and the sorting is not optional anymore.

Not sure which half of your archive is carrying you? The readiness check is built to sort exactly that.

Go and do something this week that nobody else in your niche has done, then write it up. That has always been the good version of this job. It is now also the defensible one.