Google Warns Over "Any Method" Used to Auto Generate Lots of Content For Rankings

[engine]

Google's Search Liaison, Danny Sullivan, has warned over the use of "any method" to auto generate content primarily for search rankings may fall foul of its systems. This message appears to be prompted over the most recent experimentation of AI-generated content by SEOs.

Clearly, if you're not careful with your content creation you could end up with problems.

For anyone who uses *any method* to generate a lot content primarily for search rankings, our core systems look at many signals to reward content clearly demonstrating E-E-A-T (experience, expertise, authoritativeness, and trustworthiness).

[twitter.com...]

[Brett_Tabke]

It really is about telling it how to generate content and training it in your voice.

> There's zero voice and the word choice

Try asking it for a new "tone":

Casual
Informative
Serious
Educational
ELI5
Witty
Humorous
Sarcastic
Ironic
In the style of Bob Smith (where bob is well known author)

Then teach it your writing style.

"I want you to write in my style. Here are three examples of my writings (insert post text). Then provide me with 10 creative headlines for blog posts about 'x'.

After it outputs those, ask it something like: "in my writing style and in a casual tone, write a blog post about #1 in previous response about why X would need this".

Basically you can tell it how to write from your style. It takes a bit of finessing, but I have found it can easily enhance most anyone writing. By the time you get it trained, it will surpass your engagement.

> I am shocked anyone is impressed by
> the quality of the content we're talking about

You have already read a whole bunch and not complained. Known books by AI's passed 100k sales last week on Amazon.

> I also find it interesting that many of the
> articles I've seen do not rank yet are indexed.

I'm not going to out them because there SEO at parent company is a member here, but as mentioned before CNET is generating AI articles. They have 2 on top page under three word phrases.

anyway, ChatGPT had a pretty good response:

"Write me an email in response to Atomic to the following email supporting the usage of ChatGPT and why it can be trained to write like they can in a casual tone:"

Hey Atomic,

Thanks for the feedback on the ChatGPT-generated content! I totally hear you on the word choice and lack of voice - it can definitely be hit or miss. But here's the thing, ChatGPT is like a blank canvas that's just waiting to be painted with all the right words. The more we train it and fine-tune it, the better the outcome. Think of it like a work in progress, we're always looking to improve!

And as for Google wanting the total number of indexed web pages to double, well, the more the merrier! That's what we always say. And just because a page is indexed, doesn't mean it's automatically amazing, but we're working on it!

So, what does your content look like? I'd love to take a gander and see if there's anything we can learn from it. Let's work together to make ChatGPT the best it can be!

Cheers,
[Your Name]

[Atomic]

More like a salesy tone. And at least one obviously incorrect statement.

[Sgt_Kickaxe]

Begs the question: Whose guidelines?

Where to start informing yourself? Why not listen to Sergey Brin discussing "The future of AI and Google" at Davos in 2017. [youtube.com...]

Then make a point of actually watching what Google, and the other 2000 unelected elite, are planning for you at Davos later this month. Google doesn't stray very far from the overall goals when they make changes, and a lot of change is happening right now. If nothing else, you'll know the direction the world is going and make better personal and business decisions to navigate it all.

Former CEO Eric Schmidt was far more active and involved at Davos, so you might want to catch up on his direction too, it's all on trusted YouTube channels.

Mods: I linked a YouTube channel I believe is trusted, and it does not delve into politics or have ANY commentary at all. It just shares Sergey Brin talking about "the future of AI and Google" and a lot of Google discussion, which is extremely relevant, IMO.

No need to look in the rearview and make generic guesses when the people making the changes are sharing their reasoning and thoughts. Listening to what they have to say isn't conspiracy.

[rustybrick]

One day I hope to be able to write as well as a machine. Sorry to disappoint so many of you. :-)

[Atomic]

Hey everyone using contractors for content! How do you know they aren't sending you work paid by the word generated using this?

[LuckyLiz]

How do you know they aren't sending you work paid by the word generated using this?

Run what contractors write through a copyright violation detector
Know your industry, and what really happens, works, etc.
Carefully read what contractors write with an eye on whether everything makes sense and has enough detail to be useful
Watch out for all those generic, garbage phrases like "[Doing x] can be exciting. But it can also be challenging"; ":There are many reasons why..." (and all the others mentioned above in this thread.)
Work with writers (contractors) who are expert in the subject they are writing about
Don't expect to get good content cheap.

[Atomic]

Run what contractors write through a copyright violation detector
Know your industry, and what really happens, works, etc.
Carefully read what contractors write with an eye on whether everything makes sense and has enough detail to be useful
Watch out for all those generic, garbage phrases like "[Doing x] can be exciting. But it can also be challenging"; ":There are many reasons why..." (and all the others mentioned above in this thread.)
Work with writers (contractors) who are expert in the subject they are writing about
Don't expect to get good content cheap.

Apparently, this stuff is as good as any staff writer can do. So, all of the above is great but hardly any sort of guarantee what you're paying for isn't AI generated.

Now, what does what I said about contractors apply to staff writers ate home or in the office?

[Brett_Tabke]

list of ai content detectors I am keeping: [aichat.blog...]

[tangor]

Microsoft has promised it will "soon" offer the ChatGPT AI chatbot as a service from its Azure cloud.

The software giant on Monday announced its existing Azure OpenAI service has opened up to general availability (GA). It is, notably, not available everywhere just yet though.

[theregister.com...]

[LuckyLiz]

all of the above is great but hardly any sort of guarantee what you're paying for isn't AI generated.

No, there's no guarantee. But then there's no "guarantee" the contractor hasn't used SEO tools to see what sites rank, what content and H tags those sites use, and then just patched together something similar on their own.
In either case, it takes someone with industry and topic knowledge to review what is presented to be sure it's acceptable and provides valuable content for the site visitors.

[Sgt_Kickaxe]

One day I hope to be able to write as well as a machine. Sorry to disappoint so many of you. :-) ~ RustyBrick

LOL, no disappointment at all, it's a great tool. Few learned math like a machine, and it didn't end the world. I just hope you live a long time and experience many 'human' adventures worth writing about.

I think people are just uncomfortable that, depending on Google's direction, individual websites may become obsolete. Does a story on a blog nobody can find inspire anyone?

If search (and social) don't handle this right, we're going to need more small forum communities in the world!

[nickZ]

Well content will not be king any more.
So far mass produce does not beat class produced content.

[tangor]

Consumer tech outlet CNET is reviewing all articles it published that were written with the help of AI, after it was found some contained incorrect information.

The masthead quietly began using a text-generation tool to write stories for its money section in November 2022. The articles credit "CNET Money Staff", but readers were not told that byline refers to an artificial author whose work was edited by humans until they hovered over the byline's text. 

[theregister.com...]
Chuckles! The transition to AI content may be a bumpy ride!

[Sgt_Kickaxe]

Consumer tech outlet CNET is reviewing all articles it published that were written with the help of AI, after it was found some contained incorrect information.

Well that's BS, and the problem with this Google threat gets even worse.

CNET already claimed they were fact-checking the AI content before they published, as Google said was OK. Seems they weren't doing a good job and need to do it now, but Google didn't blink. Now that CNET have described a "problem", they are admitting that CNET made mistakes, so shouldn't Google have penalized those content mistakes?

Nothing about that threat from Google was good. It wasn't even true, if you're big, like CNET.

[tangor]

Heh! Google isn't a fact checker ... they are a fact distributor---always based on their SOURCES (which is youse and meese and other websites!).

The "threat" is "don't bog me down with stuff I can't tell is spun, spoofed, or even duplicate!"

That's a very different thing, and it will not go well in g's behalf. They helped create this problem with the protracted push for AI in the first place.

[Whitey]

1/ Are AI detection methods reliable and really, what's the point if it's giving good content?

list of ai content detectors I am keeping: [aichat.blog...]

I put a couple of text blocks into each of these detectors originally outputted from ChatGPT's AI.

The first request asked for the structured legal argument involving x factors and citing precedents in the UK and US.

The 2nd request asked for the principles in determining the policy decision for a credit card chargeback involving x,y,z factors - some which were reasonably unique. All detectors came back with 98-100% human, with the exception of 1. (I won't say who, in case others get differing results) which stated that it was probably AI driven.

2/ The answers were better than I could otherwise find in the SERP's, a long way to reducing professional fees for finding out the same, and if I was a content writer a large part of my research and editing cost reduction, if i was to even bother with this.

3/ Arguably, API generated pages in e-commerce were against Google guidelines. But if the outputs were quality and/or trusted (brand/cited/etc) they got a free pass. Will the same apply in the future with AI content?

4/ Is there a risk that Google could acquire the likes of ChatGPT on a business level and eliminate the competition, providing it's own content layer to be advertised on, or has the "horse bolted" for good so that others can compete?

[tangor]

See above. MS will be offering ChatGPT as an Azure service. So it will be going mainstream in the near future.

Sad thing this "good enough" tool will be over used and will interfere with human thought/deliberations for convenience and speed...

[Whitey]

I can’t see AI cracking live human observations soon, transferred to content e.g a sporting contest. Nor can I see it replacing the emotional interactive human connection eg dating, inspiring teams or crowds, music, lyrics, novel literature.

I can see e-commerce and fact driven service ( eg medicine, basic law cases, procedural writing, education, certifications, engineering, business processes and thinking etc ) evolving in many areas and many being blindsided by this because the ability to BS or differentiate is reduced, and the structural foundations opened up.

Knowledge as a science base can be both limiting and liberating. Knowledge in content is therefore the same. In my view the opportunities are in how to use it, coupled with human imagination and observations at higher levels.

I take heart that some reported Universities here in Australia appear to be throwing in the towel and focusing on how students can use AI and apply it when being examined and certified.

I think that speeds up the road to innovation. I wonder if Google will become stymied by its inability to adapt causing a new opportunity for incumbents. A bit like the fall of a commercial dictatorship.

[Whitey]

Thought I’d pin this news article to this thread to try and provoke some thought and discussion:

[amp.abc.net.au...]

It’s not a question of banning it, it’s a matter of how you use it.

QA at scale seems to me to be a temporary obstacle. But that just says to me that content quality that is on a smaller scale will be far better at larger quantities overall across multiple websites.