What does Google think of AI content?

Want to stir up a controversy on SEO Twitter?

Ask if a website with AI-generated text can rank on Google and watch the fireworks ensue.

This is a hot button issue lately, though it’s not a new issue. Auto-generated text has been a thing for the past few years and it’s been a point of contention in the SEO community.

After all Google has held firm that AI-generated text is a big no-no for SEO. But is that still the case given how quickly this industry has expanded in the past two months?

Since the advent of ChatGPT, Bing’s AI search, and Google Bard, Google has swiftly (and quite drastically) changed their tune when it comes to AI content.

So let’s settle the debate: can you rank with AI-generated text or not? Will Google penalize your website? And how can you implement AI-generated text in your own SEO strategy?

This is a quickly evolving field. The information we have today could be obsolete a year from now, or even a month from now. Still, I’ll do my best in this article to give you a plan of attack when it comes to using AI-text in your SEO strategy.

So.. how do ChatGPT and other AI models generate their text?

ChatGPT is the most famous right now, but these AI models all work in a relatively similar way.

The AI model is trained on a massive dataset of text. It's not nearly as much as is indexed by Google, but it's a fairly substantial amount of text across industries and disciplines. At a very basic level, the model is trying to guess what the response ought to be and what each word following the previous ought to be.

Previous to GPT3, the results were unimpressive. The grammar was extremely poor, and the actual information in the content was convoluted and often incorrect. Now though? OpenAI is getting better at guessing what the next word should be grammatically, but there is still no base awareness about what the output is. That means that ChatGPT will give you a coherent answer, but it will not "know" what it is saying and be able to validate whether its guess is correct.

What does Google think about AI content?

Up until February 2023, Google’s official stance was to look down on AI-generated content.

According to Google, auto-generated text was inherently not useful, and would be penalized if caught. It wasn’t stated explicitly, but that was the undertone of the message.

Google wasn’t wrong. AI-generated text was pretty much useless for years. Even when, by a miracle, it was grammatically correct, it was still only providing surface-level information at best and riddled with incontinuities and misinformation at worst.

Times have changed. While misinformation is still an issue, it’s come along way and for the most part the outputs are surprisingly well-written and helpful.

It seems Google has caught up with the times.

Google’s new stance is to recommend webmasters to prioritize experience, no matter how the content was generated. This is part of Google’s new emphasis on EEAT - Experience, Expertise, Authority, and Trust.

For now, it seems safe to say that Google has warmed up to the idea of AI-generated content. That does not mean you have a hall pass to generate 500 blog posts tomorrow though.

Can I rank with AI-generated content?

Want some more Twitter drama?

Look for a case study from a website benefiting from AI-generated text and hop into the relies.

Most of these tweets will feature a big, shiny up-and-to-the-right chart touting impressive results (and sometimes a course). The usual MO? Publishing content at scale to rapidly increase the queries the website ranks for.

This might work in the short term, after all the more content on your site the more queries you can target. But this is hardly a sustainable strategy.

For one, it’s not a strategy at all. Strategy involves making choices, but publishing at scale just because you can is choosing not to make a choice. This can result in short-term gains, but even if the text were human-written from a content farm I doubt it would perform well in the long run.

More importantly though, once everyone starts leveraging the same tactics, using the same AI, and generating the same content… none of it will rank and it’ll result in a race to the bottom instead of to the top of search results.

Having a well-articulated strategy for your site that aims to serve your readers though? I believe, and we are testing, that AI-generated text can help there.

Will I be penalized for publishing AI-generated content?

If this post were written 3 years ago, the answer would be most likely. Now? It’s anyone’s guess.

Publishing low-effort and low-quality content that does not adequately serve the reader and differentiate itself from other pages already ranking is hardly a recipe for success. Yet that is what most people will likely do with AI-generated text, because that is what they are doing with human-generated text.

The only thing that has changed is that Google will not discriminate content based on how they believe it was generated. At least, that’s what they say.

If that content is poor, thin, and does not serve the reader it will not perform. That is the case for AI-generated and human-generated text. Just because a person wrote a blog post does not mean the content is actually valuable. Writing is hard, and writing helpful content is even harder.

Before you use this as a license to generate 500 blog posts in the next 30 minutes, hold on for a minute. Just because these text models have advanced to writing coherently with good grammar does not mean their output is error-free. In fact, there are numerous examples (and snafus) where publishers relied a bit too heavily on AI-generated text to disastrous results. Yes the output sounded good, but it was filled with errors and inaccuracies rendering the text useless at best.

Depending on the industry you work in, misinformation in your content could result in a penalty from Google (especially for YMYL queries). If you are concerned about a penalty, try to focus less on how the text is being written and more on the value that the text provides and the experience that it creates.

How you can safely use AI-generated text in your SEO strategy

The principles of SEO have not changed, just the speed of execution.

This is good news for SEOs! It means we can fail faster and adapt quicker.

Failing fast is a principle in software development. It means shipping something quickly to learn from users - the faster you learn from your users the better your product will be in the long run.

I like to apply this principle to SEO, and it means we can fail within days… but that also means you can succeed in days when you get things right.

Rather than spend months agonizing over a content campaign, you can spend weeks on executing and analyzing the result. If you're ambitious, you can shorten that down to one week.  

For now, it seems that AI-generated text gets a pass. That means you can use it in your content development process to complement (or supplement) a human writer. You still need to make sure that you understand who your audience is, you have a handle on the questions they are asking, and that you are creating content of value to them throughout their buyers’ journey.

As always, 90% of your success in SEO lies in your preparation the rest of execution.

We’ve developed a framework for building a creative and effective SEO strategy. SEO is more than picking keywords from a keyword research tool. It requires making decisions around your customers, understanding how they use search engines to answer their questions and buy products, and mapping out how to get your website in front of them at every stage.

It sounds intense, but we developed a 4-step framework that you can follow to build this strategy. Click the button below to download the free eBook and build your plan for SEO dominance today.