The Problem of Blog Posts and AI

[Jason]

If everyone is using these tools, then the content has to be on multiple blogs. In that case, obviously it's not good for SEO. Exactly, how unique are such articles?

 

[Arantor]

Not especially, because AI will inevitably swing towards producing the statistically most common words for a given intent and outcome.

But that’s not even the worst of it: there are two much larger problems.

Firstly, volume. You as a human cannot keep up with the volume of content an AI can produce. We’re going to drown in it soon enough - Kindle already is.

Secondly, hallucinations - because there’s no actual understanding, just statistical repetition of what it already knows/has/was trained on, it will often create things that *look* correct but aren’t. We’ve seen lawyers get bashed for their filings that refer to precedents that don’t exist (but are in the right format). We’ve seen a guy be told by ChatGPT that he’d passed away and it gave him a link to the Guardian website for supposedly for his obituary - it was the correct structure/format for a Guardian obit, but it didn’t exist and hadn’t ever existed.

So not only are we going to drown in information, it’s likely to be misinformation at the same time.

 

[Jason]

Not especially, because AI will inevitably swing towards producing the statistically most common words for a given intent and outcome.

But that’s not even the worst of it: there are two much larger problems.

Firstly, volume. You as a human cannot keep up with the volume of content an AI can produce. We’re going to drown in it soon enough - Kindle already is.

Secondly, hallucinations - because there’s no actual understanding, just statistical repetition of what it already knows/has/was trained on, it will often create things that *look* correct but aren’t. We’ve seen lawyers get bashed for their filings that refer to precedents that don’t exist (but are in the right format). We’ve seen a guy be told by ChatGPT that he’d passed away and it gave him a link to the Guardian website for supposedly for his obituary - it was the correct structure/format for a Guardian obit, but it didn’t exist and hadn’t ever existed.

So not only are we going to drown in information, it’s likely to be misinformation at the same time.

I take it it's not good for SEO.

 

[Arantor]

Um. OK, let me put it this way.

SEO is competing for views on a few hundred competing articles on the same subject, realistically, at any given time.

Now imagine you're competing for views when there's 10 million new articles on the subject produced in the last month.

Who cares about SEO at that point? It was hard enough finding one in a hundred or one in a thousand, but one in ten million?