With the proliferation of AI content creation tools in the past few months, more and more people are turning to tools like OpenAI and ChatGPT to generate content. I wrote an article about how to unlock the power of AI for use in your ecommerce shop. It seems like everywhere you turn, people are now talking about AI.

As such, there has been a lot of speculation as to whether or not Google will punish sites using AI to create content.

Google: We Have Algorithms To Detect & Demote AI Altered Plagiarized Content

Google doesn’t want your AI-generated SEO spam content. As recently as April of 2022, Google said, “In general, sites with spammy scraped content violate our spam policy, and our algorithms do a pretty good job of demoting them in search results.” AI-generated content is ‘spam’ and is against Google’s Webmaster Guidelines.

The irony of this is that Google proclaims to be an AI-first company. They use AI in search, YouTube, gmail, and ads because it helps improve the product.

Google is not against AI generated content and text any longer

As of January 11, 2023, SEO.ai reports that Google no longer states they are against all automatically generated content including AI generation. Instead, they specify what to avoid and what will get penalties.

The guidelines now say to avoid “text generated through automated processes without regard for quality or user experience.” Danny Sullivan, Public Liaison for Search at Google, made it clear in November 2022 that it’s not against Google guidelines to use AI to create content, as long as the output is written for people and not just the search engines.

In short, Google doesn’t care if the content is created BY people, as long as it’s created FOR people.

It is well-known that Google has a track record of taking action against websites that feature low-quality or automatically generated content. This is because many SEO professionals have attempted to manipulate Google’s ranking algorithm for their own gain.

So it made sense that Google would penalize sites for using spammy, auto-generated content. But the algorithms have evolved.

  • GPT-1 was released in 2018 and was trained on 7000 unpublished books (and the model had 117M parameters)
  • GPT-2 was released in February of 2019 and was trained on 40GB of text data from over 8 million documents. This new software model had 1.5 billion parameters. This was approximately 10 times its predecessor.
  • GPT-3 was released on June 11, 2020. It was trained on 45 TB text data from multiple sources which include Wikipedia and books. this model now has 175 billion parameters, which is approximately 100 times its predecessor.
  • GPT-3.5 is the most current version of the AI model and was released on March 15, 2022,. OpenAI made available new versions of GPT-3 and Codex in its API with edit and insert capabilities under the names “text-davinci-003” and “code-davinci-002”. These models were described as more capable than previous versions and were trained on data up to June 2021.
  • GPT-4 has only been announced recently by OpenAI and there is no fixed date of its roll-out (likely in early 2023). GPT-4 will have 100 trillion parameters, as opposed to the 175 billion parameters that GPT-3 is currently trained on. This is approximately the same as the number of synapses or connections in the human brain.

The point is, AI models are now generating really good content. Often it is impossible to determine if the copy was generated by a human or a machine. There are sites out there that make an attempt, such as Content At Scale and GPT Zero. I tested this by asking ChatGPT to write a product description (in this case for a Navajo Eagle Dancer Kachina doll). I pasted the copy into Content At Scale and it immediately detected it as being generated by AI. However, I then edited a single sentence to make it sound more personal, and tested it again – this time it came back saying that the copy was 98% human generated. I added another sentence to it, leaving the bulk of the copy as it was, and got a 100% score as being human generated. GPT Zero didn’t fair any better – I fed it an article that I wrote completely by myself and if still flagged a few sentences as being written by A.I. so, these tools are no more perfect than the A.I. itself.

Still, this is why I recommend that you don’t use AI generated content directly as is without editing it. AI can save you a lot of time, but you do need to edit the results. Not because of Google, but because you actually want your copy to sound personal. If you are looking to use ChatGPT for SEO purposes, Ann Smarty recently published an article in PracticalEcommerce which gives you specific pointers on how you can do this.

Ultimately, it doesn’t matter if the content is written by a human or a robot. Google can detect whether it is high-quality, low-quality, or outright spam. At some point, it’s highly likely that Google’s Page 1 results will be filled with content generated by robots.