AI content theft is slowly but surely killing the internet. It may take years, but ‘slow boil’ has already killed plenty of other creative industries, so don’t think it can’t happen.

The internet itself has all but killed print newspapers. Partly because, in its infancy, naive newspaper proprietors saw websites as a way to make easy money. Imagined they’d hoover up advertising, while not having the costs of newsprint to worry about.

Newspaper proprietors gave away news stories online for free. Stories they were also putting into physical newpapers they expected readers to pay for. They thought this was fine because they naively believed they had two different audiences, and would continue to do so.

Fast-forward 15 years or so and the proprietors had already got rid of every staff photographer, all the sub-editors, and; reduced the reporting staff to a handful of people, rattling around in an all-but empty newsroom. Which they then sold, desperate for short-term money.

I could also go into how the internet (free content) and camera phones have all but killed professional photography. (And AI is adding a further knife into what’s left). And how AI content is now replacing work that would have been done by PR agency writers or freelance authors.

But this article is called ‘AI content theft – killing the internet’.

Because just as streaming all but killed off DVDs and CDs (which had killed off VHS and cassettes), so AI content theft is slowly killing the thing that has all but killed off… you get the point.

Who creates internet content?

Or perhaps, I should say ‘why?’ Businesses and individuals set up websites for a couple of primary reasons. One is to make money. The other is to share their interests.

This blog is in the latter category. I like researching Cumbrian and social history. I’m a journalist, used to sharing my discoveries with others, for their benefit.

A disclaimer 

It costs me money to run the blog. There’s a host to pay, and a domain name to renew regularly (I chose the wrong one initially, but too late to change now!).

I make no money at all from Cumbrian Characters: it’s a hobby.

So if AI content theft deprives me of readers, it’s annoying, but this article isn’t ‘personal’. 

Also, to be clear, I NEVER use AI in any of my articles. I don’t need it to write and I don’t trust it to be accurate as a source.

I use em-dashes (the long ones), because they are the house style on every news title I’ve worked for.

AI content theft – hitting revenues

Google ‘Cumbrian surnames’ and the top answer is still an article I wrote some back in 2018: 
However, make it a question: ‘what are Cumbrian surnames’ and up pops an AI overview.

‘Common surnames in Cumbria include Atkinson, Thompson, Robinson, Richardson, Nicholson, and Jackson. Many of these names are descriptive, relating to specific regions or family occupations.’ 

If that answers your question, there’s no need to click on the link to my article. Or the News and Star or other news outlet articles that come underneath it.

By providing an overview, Google has reduced traffic to every site listed in its results.

That’s not going to cost me money, but many of those sites DO rely on clicks to make money.

And if it’s, say a question about which type of of paint to use on kitchen cupboards, it’s not just lost clicks/ad revenue. It’s potentially lost sales of a product.

Garbage in, garbage out

Using the example above, AI overview starts:

‘Cumbrian surnames often reflect the area’s rich history and geographical features.’

Which is basically drivel!

And Atkinson, Thompson etc are all patronyms. Nothing to do with history, geography, or, as AI overview claims, ‘specific regions or family occupations’.

Make your mind up

Ask the same question again and you’ll get a different response.

‘Cumbrian surnames, like those from many regions, often have connections to place names, occupations, or even personal characteristics’.

Totally generic: could be about anywhere. But it then suggests:

‘Allison, Booth, Blackwell, Bell…’

Well, Bell is, at least, a common Cumbrian surname. But it then says:

‘Bell likely derives from the Old English word “beall,” meaning “bell.”

Seriously?!

If you are wondering about Booth, AI overview reckons it is:

‘A common surname, often associated with the village of Booth in South West Cumbria.’

I can only think it got confused by the supermarkets!

AI overview does at least conclude:

‘AI responses may include mistakes.’

AI content theft – eating itself

Say a shop that sells fabric decides it’s cheaper to get AI to write content about ‘how to make cushion covers’ or ‘what fabric is best for bathroom curtains?’

The bot crawls the internet, scrapes information from existing websites, and cobbles together something that may or may not be accurate (depending on the sources it used).

Down the line, another fabric seller decides it’s cheaper to get AI to write content about ‘how to make cushion covers’ or ‘what fabric is best for bathroom curtains?’

And the bot scrapes content from the first site.

By now, PR agencies aren’t paying staff to research and write articles. Before they stopped doing so, they were finding the only sources available were AI-generated articles of dubious accuracy anyway. Freelance writers have long given up, their income reduced to zero.

Leaving the hobbyists. Whose accuracy is impossible to verify – think of all the disinformation ‘out there’. (For the record, Cumbrian Characters doesn’t publish anything it has any doubt about – this journalist really cares about verifiable sources and accuracy!).

And many of the hobbyists are giving up. The majority of blogs fizzle out after a few years anyway: they are time-consuming, there’s no money in it… And if AI overview is reducing traffic, then a feeling of speaking to an empty room.

Probably the only hobbyists left will be the extremists, the nutbars, the people ‘on a mission’ to spread dingbat theories and ideologies.

So there will be no fresh content being uploaded. Only AI-generated rehashes of AI-generated rehashes of AI-generated rehashes… of articles of dubious quality in the first place.

Which few will read, if they get their ‘answer’ in the AI overview anyway.

Footnote:

I read recently: 

‘It is estimated that only one in five website visitors is actually human, and over half of all internet traffic is now AI bots and crawlers, relentlessly sniffing out and ingesting human creative works.’

This seems impossible to verify and may well be a massive exaggeration. And for sure, AI Overview disagrees!

Though of course, ask it three times and you’ll get three different estimates for the accurate figures.

Meanwhile, showing how ‘useful’ bots/AI etc are:

‘Searches related to woodworm beetles: “Woodworm beetles for sale”.’

UPDATE:

Seems I’m not the only person this has occurred to!

Websites losing 4/5ths of traffic to AI overviews, survey finds