What's your take on AI generated content?

I wouldn't use AI anymore.... I am getting many pages from my website deindexed and I think it might be because of me using Jasper/Jarvis to help create the content.

But the part I don't understand is, how come the articles which are written poorly by my writers are fine but somehow the AI-written articles that provide good information and are written a lot better than crappy writers have been deindexed..

This only happens in 2 of my sites which tells me that it's only based on AI. I also have Google instant indexer thing which was also only meant for news sites or whatever but I have it on my other sites so that's not the cause of the deindexing.

I have a site that makes me a couple hundred $ which is an old site, and in there I had written 1 article with the AI and I guess it hasn't been indexed but all the other articles are fine.

FYI on GPT-2 Checkers, it gives like 0.06% fake.

I THINK Google can Check for GPT-3, Don't USE IT even as a WRITING assistant!!!!
 
Last edited:
dennis_max said:
I've been an early adopter of Copy.ai, Jarvis and a myriad of tools that generated good short-form content. Copy.ai saved me a TON of time at one of my jobs - absolutely great tool for generating thousands of FB and Google Ads.

However, all of these tool were lacking the long-form content. I've recently worked with a few developers who toyed with the idea of writing 1.5-2k articles with the AI, and the results seems promising so far. I want to build a test website to play with this idea, however, I'm concerned about Google's algorithms.

I remember reading somewhere that Google recently announced that it can detect and penalize GPT-3 type content quite easily, however, I'm having a hard time understanding how it can realistically detect that.

Overall, what's your take on GPT-3 and have you considered using it for long-form content on your projects?
Click to expand...
i disagree. they may write a synatctically and semantically correct article but people would not necessarily read it. you need that human touch. people don't even read articles written by best writers these days haha(due to video content rise), I suspect if AI could help in generating any value to business.
 
I think it is useful to generate outlines for certain types of articles.

However, for low / zero volume keywords I found it ran into difficulty quite quickly.

But it could have been that I didn't have a great grasp of the tool
 
The gpt hugging face detector out of 20 articles I tested it with caught 9/10 ai articles gave 2/10 false positives on the real content. This is impressive considering it's not even a tool from google by some independent guy. Now imagine putting 50 google engineers on this to create an ai detector, if they haven't made one already they will, and it will blow this one out of the park.

I don't mess with ai. I used to be blackhat but got tired of getting hit by every update. With all my experiences in the past 10 years of SEO everytime I found something that was a loophole that everyone knew about and was touting "the flavor of the month SEO hack" little later everyone using it got spanked.

Cloaked hacked links, BBR, senuke, xrumer, anchor text manipulation, you name it everytime some exciting way to try to screw the algo comes out you can bet your ass googles engineers are working on killing it.
 
I'm using OpenAI's GPT-3 playground, since I have access to it for free.

It's a nice complementary tool if you want to rephrase and expand some text to update old blog posts.
I've also used Rytr, but I prefer to use the vanilla's GPT-3 version, since it's more granular, and I can experiment with this technology by myself.

I have no idea if Google is able to detect GPT-3 generated content, but what matters to me is if the content is informative & unique, which I think that OpenAI's technology manages to achieve.

In my opinion, if Google detects that your blog post has a high average engagement time per user, your ranking won't be affected, even if it was AI generated.
 
I’ve used Jasper for a full month plus trial period. And I can tell you this tool is amazing for:

  • Product descriptions
  • Ads
  • Social media post captions
  • Paraphrasing
  • Creative writing

Total Inaccuracy​

But when it comes to writing an informational article, this tool fails miserably.

For the sake of argument, let’s say that the easiest way to write info articles is to write about topics you are familiar with. The other, much harder way, is to do extensive research.

When I tried to write an article about a topic I’m familiar with, Jasper just throws a bunch of false information. I could go over a few mistakes here and there, but this was the next level of inaccuracy. And I don’t mean most information was false – everything was wrong.

Now you might say it’s because of some obscure niche so AI can’t find the right information online. Nope. The search volume and SERP results are about the same as for the iPhone.

Now, you might be wondering, just as I was, maybe we can write about simple things and publish 100+ short articles monthly. Yes, AI can write “good enough” articles for simple, broad topics. For example, with a bit of guidance, Jasper can write a good article about the “blue screen of death” or “why the grass is green.”

But even here you’ll need to perform some fact-checking and with that in mind, I can also write an article about why the grass is green at about the same time.

If you count in "time on page" as a ranking factor, including site authority, I wouldn't rely on any AI tool to help me here.

Maybe there is a chance that for some uncompetitive niches you’ll find it a good fit, but I didn’t.

Article Outlines Sucks​

Yes, Jasper can create the article outline with one click. But the result of the outline flow is an unreadable article. It even totally misses the topic. No matter how much information you give upfront. If you combine it with SurferSEO, things get much worse. You might get an idea for a few headers, but in my opinion, these outlines are unusable.

Some people (a lot) talk so highly about this tool I’ve thought there might be something wrong with me. So I even try to train myself to write better guides for the tool. It didn’t help. You have to force it so much that it doesn’t make sense.

What About Writer’s Block?​

I think this is the main selling point of any AI writing tool. I still use it from time to time. When I get stuck, I just open the free Quillbot extension for Google Docs, write WHATEVER comes to mind, and paraphrase the paragraph. That’s enough to get me going.

Jasper is great for creative writing​

When it comes to writing stories, Jasper rocks. If I would like to write feature articles or stories I would definitely choose this tool as a help. For example, I’ve tried to write an informational how-to article in a story format. It was a pleasure to read. But when we take search intent in mind, I wouldn’t go this route.

Jasper is the best for writing fiction​

I’ve never gone this far, but I’m quite confident I could write a 60-80k words fiction book in a week or so with the help of this. All you have to do is think up a good plot, add characters and guide the tool here and there. Again, I’m really amazed at how this technology invents stories.

How Google Treats AI Articles?​

I really don’t know, some say Google totally deindexed their AI posts, while others swear there’s no difference in comparison with human-written text. While there’s no doubt the technology will develop further, I am quite confident that Google will follow.

Conclusion​

I had some big hopes when these tools came out. Now I’m glad they didn’t work for me. I’ve invested $150 for Jasper and SurferSEO combo and some crazy amount of working hours just to learn how to use these tools, so I could produce articles faster.

In retrospect, it would be a much wiser decision to invest those 200+ working hours and $150 into some writing classes and reading some books.

With that in mind, I think it makes much more sense to invest time and money into personal writing skills and developing processes for hiring content writers if we talk about content websites.

As for the writing assistants, I wouldn’t go further than Grammarly and Hemmingway App.
 
Hi all, I have several blogs on which I conduct experiments.
In my experience, AI-generated content is well indexed and ranked.
If this content is of high quality, then there are no problems with it.
 
AI content generators are just a tool, and like any tool it's how you use it that counts. I use open source and alternative AI APIs (GPT-NeoX, AI21 etc) and find them to be great as a starting point for articles on subjects I don't know much about. With automation and a bit of automatic filtering they're fine for lower tier link building too in my experience.
 
I know this is an SEO forum and all that, but I think there could be some fantastic benefits to AI writing outside of the SEO world someone could take advantage of. Use SEO to draw in clients, then use the AI tools to help you along the way. Stuff that doesn't need to be some grand masterpiece.

Protip: treat the tool like an intern/VA. You wouldn't have them go buck wild; you give them specific tasks and explain exactly how to do each task.

Resume writing service
Really easy. With just a basic understanding of a person's role, you should be able to tell if the stuff it's spitting out is good or not.

Or, you could do what most people aren't smart enough to do and just copy someone else's job description ... then run it through a rewriter.

Boom, charge a few hundred bucks and tell them to tell their friends.

Business paper writing service(s)
We're talking whitepapers, marketing and business plans, directory listings of any kind (real estate, anyone?), or any sort of circle-jerky content that SMBs love.

Fed Jasper a few key points from a website I sold, and it gave me:

Founded in 2019, [Website] is a job board catering to the [industry] industries. In July 2021, the company was sold for an undisclosed sum. It grew to over 12,000 unique visitors monthly in fourteen months. The company generated over $54,800 in total value/revenue, with $4,300 spent on content creation.

From 2019 to 2020, [Website]'s revenue grew by 66%. The company's sale is a testament to the growing demand for digital tools and resources in the [industry] industries. [Website]'s success underscores the need for niche job boards that can provide targeted leads for specific industries.

A few minor tweaks and (if the numbers were actually impressive) I have the executive summary for a client case study created.

Email marketing DFY service
I mean, shit, look at some of these examples: https://flowium.com/copy/

Do that for $250+ for each email created, find a third-world designer to do that part of the process, and boom.

Money.

I'm using it to write emails now. Write a key sentence and let it expand for two sentences. Repeat until finished.

Don't have to worry about not being indexed lol
 
freshpeppermint said:
I wouldn't use AI anymore....
Click to expand...
I think the idea of using AI to bulk publish tons of articles targetting zero comp long-tail keywords is so seducing.

It was so seducing that I went balls to the wall and punched out 2000 posts with AI.

What were the results?

I'd say 50% decent and 50% pointless.

Overall, I think it makes more sense to just focus on writing good human content.

That said, if you are just trying to rank for clients or rank n rent then you could leverage AI to take the top spot but for a blog, I think good content will always win.
 
Weird question but asking because: I have been experimenting going hard with AI

And inevitably, the articles are short, direct and answering very straight-forward, zero comp search terms.

Would it be better to just pile all the AI articles into one juggernaut or leave them as short, separate articles?
 
Can we monetize the affiliate program just by creating content with AI?

Have you tried this before.
Thank you.
 
“AI” content is useful, I use Frase to write product descriptions and get page title ideas, etc. I’ve tried to use it for long-form writing but found that with the amount of editing and fact-checking it's easier to write content myself.

I'm interested to know how this upcoming Helpful Content Update will affect “AI” content. Google specifically mentions “automated content” in their list of what unhelpful content is.

I don't think Google is as advanced as they like to think. In general their results have been mediocre for a couple of years now since they started relying more on machine learning.

Some SEOs are asking why they announced this update with so much detail beforehand, was it to scare webmasters into reducing the amount of optimization they do on content to rank it because Google’s algorithm favors optimized content?

I’m hoping this update will result in more business (content audits) like Panda did :smile:
 
Hoping to add my 2 cents in here.

I'm using GTP-3 content for brand new sites, targeting 0-10 MSV long tail keywords. I write an outline of the H1, H2s, and H3s that we want and fill in the blanks with PAA things that appear when searching those 0-10MSV keywords.

I'll build out a few different silos around big keywords I want to eventually target, and fill them in with a hub page plus 5-6 sub-pages each.

While that's in motion, I follow the OG post from @CCarter on WickedFire, the Big Brand Playbook. Obviously modernized for 2022 but keep following.

I'll build legit citations, nothing crazy. Once I see content start indexing (and ranking) in top 50 results, I'll toss a few legit links sprinkled between home page, hub pages, and sub-pages targeting those 0-10 MSV keywords.

Turns out, that's enough to push top 3 results for those low keywords.

You might be thinking, "But why target such low volume keywords?"

Well ccarter explains it all perfectly in his Avalanche Method post. I have no search volume, those are the keywords I can write and rank for. And usually there are enough variations of the long-tail phrase that are also 0-10 KWs that I can rank for those on the same pages.

So now I have 60 keywords getting me 3-7 users per month. That's 180-400 visitors per month, usually within 90 day, and I'm supposedly still in the "Sandbox".

I repeat this process until I have 100+ keywords bringing me at least 500 visitors per month. THEN I hire writers to write new, better content for the hub pages, and for all future content. Lather rinse and repeat, pick keywords with slightly higher search volume, get it written by real writers, do good on-site SEO, silo structuring, etc. and the site can take off.

Is it a lot of work to make those initial posts? Not really. I can literally just make an outline of a single H1, three H2s, and 2 sub H3s before the final call to action paragraph. GTP-3 does all of the work to fill in the gaps. I edit it in 5 minutes with Grammarly and throw it into a tool like Surfer for a final tweak. All in all I can write a new page in 10-15 minutes and have trained my VAs to do the same.

Here's a site I launched in November that's done 14,000 visitors in the last 30 days (12,000 from search, another 2k from ou email list):

BQu6T4l.png

GlvyKUF.png



This is the exact method we followed. Some of that GTP-3 content now ranks for much bigger partial keywords than the long-tail we were initially targeting. But it's supported by steady growth, daily posting, and begin being heavily supplemented with backlinked in early April (picked up a massive, juicy footer link from a huge topical authority website to a calculator tool that we created).

Our biggest keyword in spot #1 right now is a 2.4K volume keyword for that tool. We have 19 total keywords in the top 3 for that page, and we're closing in on another keyword for this page that will be 3.2k volume (currently rank 5).

oYcUmBO.png


Yeah, we custom coded a calculator / estimator tool for this niche which users find incredibly valuable, but every single word on the page was generated by GTP-3.

Have repeated this process with 3 other sites in 2022 (4 total now).

Site #2 - Only has 8 pages right now

8x31n9Q.png


Site #3 - This site was launched on April 1st and has generated over 80 leads so far for a roof replacement contractor in a top 10 US city market. Same strategy, uploaded initial pages with GTP-3 content, have now switched to real writers. Check out that bump

4LMOMWp.png


So does this work if EVERY PAGE is GTP-3? Probably not. I really don't know and don't want to waste the time to find out. But this strategy of seeding the site with GTP-3 content going after 0-10 MSV phrases until we have some decent traffic to go after bigger terms, and then swapping to real writers for the meat and potatoes of the remainder of the site, seems to work for now.
 
AI content is getting quite good with custom trained models. However with public ones like Jasper and similar tools you won't get too much especially now with the new "helpful content update". Don't get me wrong even articles from those public tools will pass google checks as long as they are properly edited and while you don't let AI write a whole article unsupervised.
 
PuppetMaster said:
So does this work if EVERY PAGE is GTP-3? Probably not. I really don't know and don't want to waste the time to find out. But this strategy of seeding the site with GTP-3 content going after 0-10 MSV phrases until we have some decent traffic to go after bigger terms, and then swapping to real writers for the meat and potatoes of the remainder of the site, seems to work for now.
Click to expand...

In my humble opinion, right now you are riding a freshness boost. The fact that you're at DR 7 and your sites are new, is a clear indicator of that. New sites that are pumping content constantly usually experience an increase in rankings in the first months from the "freshness" score.

I had the same experience with new sites that were ranking for BIG search volume (like 40k+ search volume) even though the DR was MUCH lower than the bigger outfits. I always thought it was because our content is great (which it is) but after a few updates - rankings will drop and they will put you in your "rightful" place (as one of my friends would say).

I am just telling you this incase this rude awakening happens to you. Be prepared for it. I am not using any AI content and was still nervous when the drops happened.
 
Hi, i've seen a loot of topics in regards of AI Content Generator, what is your point on this? does anyone trying this kinda tools and "success" to build some site out of that content? ofcourse despite a bunch of seo stuff regards on domain metrics, aged, etc.

Thank you.
 
ChatGPT / AI content seems to be literally blowing up right now, in the last few weeks. I have been many, many SEOs talking about it and posting various examples.

Anyone else concerned that the serps are going to get completely overwhelmed with spam/non-human content?

I mean, this stuff it very hard to detect. I assume people here are already pushing out millions of pages of content.

For those of us that employ humans to write though, what are the consequences? Sure, our content will probably still be marginally (or greater) better, but still?

Is the new strategy not to train writers to write content, but to create excellent ChatGPT prompts?
 
Darth said:
Is the new strategy not to train writers to write content, but to create excellent ChatGPT prompts?
Click to expand...
I think the strategy is to use GPT-3 content to collect data via Search Console, and use that data to get posts written by humans.
 
Darth said:
ChatGPT / AI content seems to be literally blowing up right now, in the last few weeks. I have been many, many SEOs talking about it and posting various examples.

Anyone else concerned that the serps are going to get completely overwhelmed with spam/non-human content?

I mean, this stuff it very hard to detect. I assume people here are already pushing out millions of pages of content.

For those of us that employ humans to write though, what are the consequences? Sure, our content will probably still be marginally (or greater) better, but still?

Is the new strategy not to train writers to write content, but to create excellent ChatGPT prompts?
Click to expand...

Hard to know but a few things off the top of my head that come to mind:
  1. I can see AI content flooding the market until eventually the mainstream becomes aware that a ton of content they are reading, when doing google searches, is actually written by AI, not humans. When it gets to that point, it will be in the news, cartoons (like South Park had the clickbait articles loaded with ads), etc. And google will need to do something about it publicly. They will come down hard. There will be a reckoning and many non-ai content sites will get smashed in the process. I imagine they will again strengthen the power of backlinks. And possibly come out with a new version of EAT guidelines that sites must clearly show human authors, etc etc - but in reality what will really matter is the links. Those without the links - like so many on this site preach (I can grow my site without links!) will be starved. Basically a flood and then a famine. So it's going to come down to who can survive both - that's a long stretch of getting fucked. So better get your margins up and have cash on hand to survive. But those that do survive will be in position to take huge market share and grow exponentially.
  2. A big winner in this will be video content with actual humans in the video, not ai generated videos. And perhaps sites that have video content with humans will actually get a boost due to Google wanting to push human content more to show Google results can be trusted.
  3. If google fumbles this it leaves a huge opening for the competition to grow. Bing but perhaps Apple especially.
 
Sutra said:
Hard to know but a few things off the top of my head that come to mind:
  1. I can see AI content flooding the market until eventually the mainstream becomes aware that a ton of content they are reading, when doing google searches, is actually written by AI, not humans. When it gets to that point, it will be in the news, cartoons (like South Park had the clickbait articles loaded with ads), etc. And google will need to do something about it publicly. They will come down hard. There will be a reckoning and many non-ai content sites will get smashed in the process. I imagine they will again strengthen the power of backlinks. And possibly come out with a new version of EAT guidelines that sites must clearly show human authors, etc etc - but in reality what will really matter is the links. Those without the links - like so many on this site preach (I can grow my site without links!) will be starved. Basically a flood and then a famine. So it's going to come down to who can survive both - that's a long stretch of getting fucked. So better get your margins up and have cash on hand to survive. But those that do survive will be in position to take huge market share and grow exponentially.
  2. A big winner in this will be video content with actual humans in the video, not ai generated videos. And perhaps sites that have video content with humans will actually get a boost due to Google wanting to push human content more to show Google results can be trusted.
  3. If google fumbles this it leaves a huge opening for the competition to grow. Bing but perhaps Apple especially.
Click to expand...
I completely agree with these 3 things being the only solution.

At this point, AI is here to stay, and the only thing google can do to fight against it is limit what they index by requiring higher quality pages that feature unique images (or images), cited sources, and content that reads like it is from someone's personal experience.

They'll likely have to scale back up the power of links and EAT to further reduce the amount of spam that makes it onto the internet.

I don't think video content will be a solution because it's not much harder to build fake videos than AI content. So there's not much they can do about AI content overall but embrace it and increase the difficulty of recognizing it in the SERP.
 
From talking with some of our heavy-hitter clients, most are saying that they're not getting the traction they want from purely written AI content, and that it needs human editing to be viable.

I've found that those that said they had success with pure AI were being a bit disingenuous in that they were targeting keywords with such low competition that Google had no other choice but to rank it.

Personally, I'm seeing how I can use the various AI tools that are popping up to make my various processes more efficient. I'm not going to be posting AI content on my niche sites until there is a clear indication one way or another.

As a side note - Jon Gillham is working on a Copyscape alternative that also detects GTP-3 written content. So far, so good: https://originality.ai/
 
I've been messing around with ChatGPT from OpenAI for the past 2 days. I've managed to create 14 posts, all optimized with good keyword selection, headings, etc.

At first, I was impressed with the output. As time has gone on, however, it's become apparent that the same issues that plague Jarvis (and other tools) are still present. The exact same phrasing is used multiple times, information is repeated, etc. Oh, and all the AI content detectors catch on immediately.

All-in-all, there is no way that a website filled with AI content wouldn't get smacked down. The content quality just is not there. I know people are using these tools as helpers, but I don't think it's worth the effort to create an efficient workflow process with them.
 
PuppetMaster said:
I'm using GTP-3 content for brand new sites, targeting 0-10 MSV long tail keywords. I write an outline of the H1, H2s, and H3s that we want and fill in the blanks with PAA things that appear when searching those 0-10MSV keywords.
Click to expand...
You should do a case study or provide regular updates on this. Would be very interesting.
 
You must log in or register to reply here.
Back