zh
English Español Deutsch Français Italiano Português (Brasil) Русский 中文 日本語
Submit post
Go to Blog

Yes, An AI-Generated Blog Can Rank — And That’s Scary

85
Wow-Score
The Wow-Score shows how engaging a blog post is. It is calculated based on the correlation between users’ active reading time, their scrolling speed and the article’s length.

Yes, An AI-Generated Blog Can Rank — And That’s Scary

Kristin Tynski
Yes, An AI-Generated Blog Can Rank — And That’s Scary

A few weeks back, OpenAI finally released its full version of GPT-2, a state-of-the-art text generator they previously called “too dangerous to release publicly” back in January. 

New tools like this one mark a new era in digital content creation with massive implications for organic search and SEO in general. The ability to generate unlimited text that readers (and likely search engines) think is genuine human-written text presents a potentially existential threat to SERP quality. 

Given this, I decided to run an experiment. I created a blog entirely generated by AI called This Marketing Blog Does Not Exist.

I wondered: Could it rank? Could it drive traffic? Would it be caught and penalized? 

And most importantly: Is this a viable new blackhat technique that we will see exploited by bad actors for financial gain?

Here is What Happened

After publication, through highly-targeted digital PR outreach, we were able to secure interviews and press coverage with top-tier online publishers.

As a result, we accumulated high domain authority links right off the bat, from publications like:

These initial media stories generated decent additional syndications, as well as at least one major additional pickup months later at Venture Beat (due to publishing scheduling timelines).

We can see this accumulation of root linking domains reflected with the charts below:

Yes, An AI-Generated Blog Can Rank — And That’s Scary. Image 0

In the roughly 4-5 months since the blog was launched, it acquired links almost 220 unique linking domains, with 50+ having a domain authority of 50 or better. 

This clearly set the groundwork for relatively fast indexing of the entire site (took about a month), as well as some good early domain authority building to allow it to begin ranking in the longtail almost immediately.

Yes, An AI-Generated Blog Can Rank — And That’s Scary. Image 1

I want to note that I did no onsite optimization. I simply uploaded the text generated by the AI model and Idid no additional optimizations for specific keywords, article comprehensiveness (with Clearscope or others), title/heading optimization, meta optimization, page interlinking (except the WordPress category links in the sidebar), etc. 

Given this, the site ended up ranking for nearly 300 terms, according to SEMRush, over the 4 months since its launch. 

Total Ranking Keywords: 292
First Page Rankings: 3
Monthly Traffic from Organic: ~60 visits per month

So What Does This Mean?

It is clear that press coverage conferred a great deal of early high domain authority links, which provided the basis for the ranking ability of the ~600 pages generated. 

But because this was an entirely new domain, the press links could only take rankings so far — 215 ULDs (unique linking domains) is a great start, but it is certainly not enough to allow for many first-page rankings outside the extreme longtail.

Had I put more planning and effort into onsite SEO optimization and implemented a hub/spoke content organization, I am confident we would have seen better keyword/traffic results.

The blog’s 600 pages only took 20 minutes to generate. Adding additional pages would most likely create a corresponding increase in total ranking keywords and organic search traffic. 

Extrapolating on the results seen with 600 pages, the following could roughly be assumed.

  • Increasing to 6,000 AI-generated pages could mean:
    ~3,000 keywords ranking
    ~600 visits per month

  • Increasing to 60,000 AI-generated pages could mean:
    ~30,000 keyword rankings
    ~6,000 visits per month

  • Improving onsite SEO, site structure, internal linking, etc., could have massive effects, with exponential improvements being realized on theoretical sites of greater and greater size. 

In essence, it does appear possible for someone to create a highly visited blog in a day without a word of human-written content.

Where Do We Go From Here?

As anticipated, some companies have already begun to monetize AI-generated text for SEO. 

KafkAI is an early example; it seems likely we will see more companies trying to cash in, further disrupting the current marketplace of low to mid-quality content creation companies, article spinners, etc. 

I see the advent of this technology as a potential existential risk for web content, and Google’s ability to continue serving relevant content in the SERPS. Recent news of the incorporation of BERT (same architecture as the AI models talked about in this article) is telling news. 

Google is indeed looking to incorporate more and more state-of-the-art AI into their algorithm, but so far, it appears that it doesn’t include trying to identify or filter content created by AI text generators. 

Perhaps continued discussion around the risks to SERP quality and the potential for overall degradation of the web’s content ecosystem will spur new efforts by Google to find — and filter out — this next generation of webspam. 

Kristin Tynski
Pro

Asks great questions and provides brilliant answers.

Kristin Tynski is the SVP of Creative and a co-founder at Fractl, a boutique growth agency based in Delray Beach, FL. Prior to Fractl, she was one of the owners of Voltier Digital, a full service agency that was acquired in 2011.
Share this post
or

Comments

2000
Appreviewcity
Newcomer

Either just recently joined or is too shy to say something.

How can I write AI content free? Which AI tool you used?
Please explain me how can I write AI content?

I tried talktotransformer.*com but it's doesn't work properly.
Talktotransformer content is incomplete.
Tinkleo
Newcomer

Either just recently joined or is too shy to say something.

Is this technology mature? I don't think there is such an exaggerated AI technology, is it? Better than Google?
Rosie Pugh
Enthusiast

Occasionally takes part in conversations.

Wow, I had no idea that AI writing could read so convincingly. I hadn't expected nuances such as tone of voice, but there is a clear difference on your blog between the "chattier" tone of the social/influencer posts and the more technical style of the SEO posts.

Thanks for sharing the results of your experiment - it will be very interesting to understand how this affects content production in the future!
Helper

An experienced member who is always happy to help.

Yeah, It's real scary to search engine's to pick and deliver quality contents..
Newcomer

Either just recently joined or is too shy to say something.

Would you share some of the anLytics data? I a. Wondering if the visitors spent time on site or multiple pages. Or did they all bounce?
Newcomer

Either just recently joined or is too shy to say something.

Interesting, new ideas for my brain.
Peter Macinkovic
Newcomer

Either just recently joined or is too shy to say something.

GPT-2 is amazing and I can see how this can be implemented on scale.

Since Google currently overvalues content, one possible use case I can see is that AI generated content can be used as a stop-gap as part of as 'throw things at a wall and see if sticks' type of strategy.

If pages get traction, that can prompt the content owner to revise it with a manual touch.

If AI-generated content that is of a quality near-human readable level becomes prominent, the only response from an algorithmic standpoint is to devalue content.

I can also see how this blog taking only 20 minutes to generate and performing 'decently' from a search standpoint that this can be prime fooder for a slew of private networks being generated at a push of a button - 600 blog posts of uniquely generated content, what if 5 or so were choice backlinks to a money site?

It certainly makes the future a lot murkier.
Rahul Vij
Newcomer

Either just recently joined or is too shy to say something.

Great share, really interesting experiment

but you got lot of links saying that this is an AI generated content and it got you traction,

sub standard content on a fashion site won't attract that kind of links and won't help in getting the rankings.
Similarly extrapolating data in a linear trajectory is not exactly what will actually happen as no other pages will keep getting the kind of traction this page got. and hence such rankings and traffic will not come.

I think the ranking you got was because of the buzz around the article,which will not happen for a site trying to monetize sub standard content.
Newcomer

Either just recently joined or is too shy to say something.

now google will have a really tough time there ones again
Normand Lemieux
Helper

An experienced member who is always happy to help.

What if we were to turn in the quotations to real influencers for review/approval? What if we were to replace fake Instagram filters with existing filters?
We would get a real, legit blog post for a fraction of the effort needed for conventional blog posts.
It won't take long for such AI to take more elaborate instructions from editors.
Take heed freelancers, AI is coming for your job.
Newcomer

Either just recently joined or is too shy to say something.

Really Scary! So Where it will fall in White Hat or Grey Hat. I feel Keyword Stuffing will be a key concern here.
Sebastian Hovv
Enthusiast

Occasionally takes part in conversations.

Wow that is scary.
Agus Nasrulloh
Enthusiast

Occasionally takes part in conversations.

Interesting and useful
Newcomer

Either just recently joined or is too shy to say something.

Yeah, but it seems like you simply got backlinks by telling people the blog was AI-generated (which of course is going to get a bunch backlinks). IOW, the backlinks from The Verge, AdWeek, etc are all articles similar to this one on SEMrush.
Telson
Newcomer

Either just recently joined or is too shy to say something.

intersertingg
Mansoor Jarwar
Enthusiast

Occasionally takes part in conversations.

Good one and informative work. Thanks for sharing it with us :)
Marcus Miller
Newcomer

Either just recently joined or is too shy to say something.

Interesting. But just a new kind of spam I guess and Google will find a way to crack down on abuse. The never ending game of SEO cops and robbers or cowboys and Indians continues...
Newcomer

Either just recently joined or is too shy to say something.

Great read! I'm stunned that the site got as far as it did in such a short period of time...it's incredible.
Newcomer

Either just recently joined or is too shy to say something.

Hey Kristin, many thanks for sharing your amazing research. Though I still can't accept that any software can itself create decent quality unique content, but if it's possible then this will have a seriously detrimental impact on online content writers.
Newcomer

Either just recently joined or is too shy to say something.

What software did you use to generate this? Transformer?
Kristin Tynski
Pro

Asks great questions and provides brilliant answers.

Vikrant Sharma
Yes, specifically the Grover model
Newcomer

Either just recently joined or is too shy to say something.

Kristin Tynski
Can you share the link? What was the cost?
Kristin Tynski
Pro

Asks great questions and provides brilliant answers.

Vikrant Sharma
https://grover.allenai.org/, we got access from the developer of it to the full sized model, and set it up on our own server. There were nominal costs associated with the server, but it was less than $50 total. https://huggingface.co/ has done HUGE things for the text generation community, open sourcing easy ways to use all the major new models. This could have been done with Salesforce's CTRL model, GPT-2, and others. Grover was used because it was trained on news articles, which was more appropriate training for producing marketing blog content.
Newcomer

Either just recently joined or is too shy to say something.

Kristin Tynski
wow thanks!
Punye Aggarwal
Enthusiast

Occasionally takes part in conversations.

Hey Kristin, many thanks for sharing your amazing research. Though I still can't accept that any software can itself create decent quality unique content, but if it's possible then this will have a seriously detrimental impact on online content writers.
Kristin Tynski
Pro

Asks great questions and provides brilliant answers.

Punye Aggarwal
It can write convincing copy, quality, that's somewhat of a different story. Modern language generation models don't generally have any built in understanding of what is or isnt factually correct. That seems a ways off. Still most of the content on Thismarketingblogdoesnotexist.com are convincingly human to many readers.
Punye Aggarwal
Enthusiast

Occasionally takes part in conversations.

Kristin Tynski
That's so amazing & super-duper interesting. I mean I could never have thought that AI can be used to recreate content on any topic in such an amazing manner. As long as the content isn't stitched together or words changed for synonyms etc., I don't think Google would be able to algorithmically ascertain that it was a software that wrote the piece of content.
Punye Aggarwal
Enthusiast

Occasionally takes part in conversations.

Punye Aggarwal
I certainly knew that this was possible, but I can't imagine the kind of hard-work it would have taken for the programmers to create such a software.
Shivbhadrasinh Gohil
Enthusiast

Occasionally takes part in conversations.

Punye Aggarwal
We can never know. But there are possibility of detecting the machine written content as well.

Here's is such software.
https://grover.allenai.org/detect

I used an article of Thismarketingblogdoesntexist and it detected successfully that it's a machine written content.

Here's the screenshot: https://meetanshi.d.pr/uSIKgj
Newcomer

Either just recently joined or is too shy to say something.

I'm not sure why anyone is surprised by this. Search engines are algorithms. Yes, they have some manual manipulation - but its pretty difficult to do that at any kind of scale. Given that search engines have to use an abstraction (simplification) of how language works vs. actual language (think BERT) that means that the same processes used to evaluate content can be used to create content. As to Cam's response below, my answer would be no - depending on the sophistication of the AI being used to write the content, Google's is going to find it very difficult to determine whether or not content is human vs. AI written. In fact, one of the few advancements I expect AI to make to complicate this even further is that in time, AI should be able to emulate not only how to write content, but how to write content from a the perspective of how a particular writer would have written about the topic. Imagine, we could have new Shakespeare. It not only potentially doable, it's probable.
Kristin Tynski
Pro

Asks great questions and provides brilliant answers.

Marston Gould
Maybe not surprising, but many people have no clue AI has gotten to this level of sophistication when it comes to text generation.
Newcomer

Either just recently joined or is too shy to say something.

Kristin Tynski
Combine this with the ability to realistically depict people visually and with their own voice - now you have a computer able to create the script.
Kristin Tynski
Pro

Asks great questions and provides brilliant answers.

Marston Gould
Oh 100%, combining text generation, AI voice cloning, and Video deepfakes and you could create a pipeline for infinite text/video/audio content generation in anyone's face/voice. The future is going to be wild.
Newcomer

Either just recently joined or is too shy to say something.

Hey Kristin,

Great read! I'm stunned that the site got as far as it did in such a short period of time...it's incredible. I guess the true test of the AI-powered sites would be to see how they do long term. We have to wonder, will Google continue to reward them with high rankings or will there eventually be a repeat performance of the Google Panda update which penalizes websites with automatic text generators.

Nice job with your research, and thanks for posting!

Cam
Kristin Tynski
Pro

Asks great questions and provides brilliant answers.

Cam Russo
Thanks Cam, I would expect Google is already working on this problem, and will have future updates that penalize content identified as AI written. How long that takes, or how easily it can be subverted seems like an open question. I think we are going to see yet another SEO arms race on this front.

Send feedback

Your feedback must contain at least 3 words (10 characters).

We will only use this email to respond to you on your feedback. Privacy Policy

Thank you for your feedback!