"Google Search's guidance about AI-generated content" (8 Feb 2023)

I've already spotted one of my entire sites spun using AI and taking the feat snippet in Bing. gg
 
Darth said:
I've already spotted one of my entire sites spun using AI and taking the feat snippet in Bing. gg
Click to expand...
Google giving up on ai spam, official postulation is if you cant beat them join them by generating it yourself directly in serps.
 
I'm still of the opinion that the best thing to do is to use AI to generate articles, gather keyword data via Search Console, and start a 100% "clean" website using the gathered data.
 
Welp, time to plaster the home page and about us saying we're 100% human made.
 
JOoa0ky said:
Welp, time to plaster the home page and about us saying we're 100% human made.
Click to expand...
How many of your users visit your homepage?

For SEO purposes, I doubt marketing yourself as "human made" will make much of a difference.

In fact, I wouldn't be surprised if some users take it as an insult to their intelligence.
 
KiwiBlogGuy said:
How many of your users visit your homepage?

For SEO purposes, I doubt marketing yourself as "human made" will make much of a difference.

In fact, I wouldn't be surprised if some users take it as an insult to their intelligence.
Click to expand...
I'm deep into YMYL healthcare so I think it matters even more in my situation.
 
JOoa0ky said:
Welp, time to plaster the home page and about us saying we're 100% human made.
Click to expand...

I would definitely take a "wait and see" approach to this, because they might just be fucking with you guys to get more data on the current AI content usage. Then in the near future pull the rug from underneath anyone dumb enough to jump the gun by hitting you guys with a Google update that wipes everyone out that has AI content.

It's like putting cheese out for rats - what's their actual end goal that helps their bottomline?

Think long term, what's better for them, let spam go wild (pun intended) for real or caught all spammers in one go, then decimate them?

I would stay away from AI content since it's just the latest fad/hype for a longtime. Doesn't make sense for Google to "give up", when they can simply use it as bait to get rid of spammers once and for all.

I guess you have to ask if you are stupid enough to fall for the cheese.
 
I think it could be a "cover our asses" approach. It'd probably hurt them to on the one hand say "your use of AI content is bad and spam" but then turn around and say "we're going to use AI content because it's good and helpful"
 
Thanks for sharing. This was inevitable. If everyone begins using AI content in their writing, and if Google themselves introduces AI to deliver answers, it was only a matter of time. It would be pretty hypocritical for Google tell website owners not to use AI content, when they are using it themselves in search.
 
The thing is for half the writers you see (at least half) on LinkedIn saying they're not scared because chatGPT sucks compared to their amazing deep human writing, we've all received an article from them much worse than what chatGPT with good instructions can produce...

The question will simply become - is it easier to make money as an SEO with AI content OR by doing content that actually beats AI content rather than just hiring whatever cheap writers you can find to fire up your 500 pages of content...
 
Most ChatGPT content I've seen is still not very good.

Maybe people get bamboozled by its ability to spell correctly and use varied language with a large vocabulary. It still seems to me that it's just a glorified rewrite of Wikipedia facts in most cases with a shitload of filler and bs. You can also literally ask it to be more verbose and that just changes the signal to filler ratio, but it doesn't add more actual facts. It's very low on facts and even lower on relations to other topics.

I think it can be good at stuff like writing intros and basic conclusions and how-to stuff, but I will only use it for supplementary content, not for the meat of the content.
 
Something I see people missing sometimes is that these "AI writing" systems all need a steady stream of fresh, quality, non-AI content in order to function. Without it, they just churn and re-churn the same stuff over and over. It's like having rewrites done of rewrites ad nauseam, and everyone knows how that turns out.
 
JesseEddleman said:
Something I see people missing sometimes is that these "AI writing" systems all need a steady stream of fresh, quality, non-AI content in order to function. Without it, they just churn and re-churn the same stuff over and over. It's like having rewrites done of rewrites ad nauseam, and everyone knows how that turns out.
Click to expand...

Exactly and that's the reason why AI content will not do well, even if Google can't spot it.

It's just rehashed, with a lot of filler and no originality.

I do want to test if I can feed it a datatable and have it write from that. I doubt currently, but soon you probably can.
 
bernard said:
Most ChatGPT content I've seen is still not very good.

Maybe people get bamboozled by its ability to spell correctly and use varied language with a large vocabulary. It still seems to me that it's just a glorified rewrite of Wikipedia facts in most cases with a shitload of filler and bs
Click to expand...
Have a play with giving it more data that isn't just from obvious sources, giving it instructions on how to structure the post, what to include etc. Break it down into sections that you can 'control' it more easily over instead of working on a long post all in one. Watch a bunch of the prompt engineering videos on youtube where people are doing clever things with it (even fine tuning it with examples on the back end so they have 'their own model' to test on the playground (paid users).

Don't let bad content you've seen others produce put you off too much. Most people are using it as if it's 'magic' just giving it 300 titles and expecting a website. I'd imagine you'd get similarly poor results if you sent 300 titles and nothing else to an otherwise fine writing agency too. But if you treat it more like you're going to send 300 really good briefs (but also have studied best how to word and set those up with the model instead of a human) to a fairly decent writer you'll be a lot happier with what you get out of it.

Even if it's not there for you yet (we still have plenty of work for our human writers here at ReachCreator...) remember GPT4 will come out and it'll be a bit better, then GPT5, then some competitor... everything you learn now will set you up for the future where it will be good enough for what you need. And even right now you'll likely find some tasks for which it's perfect and saves you a lot of time.
 
You must log in or register to reply here.
Back