Google’s stance on AI-generated content

Reading time : 8 minutes

The theme of recent weeks in the digital world has certainly been the spread of AI systems applied to every sector, and SEO has been no exception. There had been a stir over the prolonged silence of Google, which seemed “displaced” by the spread of ChatGPT et similia even to the common public, and so we are not too surprised that there has been an abrupt acceleration to the plans of Mountain View, which just in these hours is launching new features for Search based on the most advanced AI technologies. Most importantly, there also comes an official stance on texts generated with the support of Artificial Intelligence, clarifying what approach will be taken by the search engine and its algorithms for the purpose of ranking such content.

What Google says about Artificial Intelligence applied to SEO and writing

In short, Google tells us that it does not matter who – or what – writes content, as long as that content is written to help people and not to manipulate search results.

The note posted on Search Central Blog, dual authored by Danny Sullivan and Chris Nelson, is explicitly called Google’s Search Guide to Artificial Intelligence-Generated Content and seeks to show how such texts “fit with our long-standing approach to showing useful content to people in Search.”

Before we get into the details of the information, we can say that there is nothing particularly “new” or surprising, at least for those who follow Google’s evolutions (and read our blog!): the references cited are the Content Update system and the December 2022 update of the guidelines for quality raters, which are the “beacon” for understanding what Google examines to assess the quality of content and also of those who actually create and sign it.

Always and forever quality content, however it is made

The premise is that Google has “long believed in the power of artificial intelligence to transform the ability to deliver useful information“, and so there is no hostile ban on leveraging technological advances to support the work of content creation and publication.

For some time, Google has been pushing the concept of “content quality“, rather than how texts are produced, is this “has helped provide users with reliable, high-quality results for years”. For example, the article continues, even a decade ago there were understandable concerns about an increase in mass-produced but still human-generated content: no one would have thought a Google ban on all human-generated content was reasonable in response, and it was much more effective and sensible to “improve our systems to reward quality content, as we have been doing.”

Focusing on identifying and rewarding in ranking quality content has been key for Google from the beginning, and this path continues to this day, including through ranking systems designed to surface reliable information, such as the aforementioned Helpful Content. Introduced last year, this algorithmic system aims to ensure that users using Google can receive in response links to content pages created primarily for people, rather than to please search engines and gain visibility positions in SERPs.

How automation can create useful content

The picture does not change even in the face of the latest developments, and Google remains consistent in its approach: anything used with the primary purpose of manipulating search result rankings is a violation of anti-spam rules and is at risk of manual action. This also includes automatically generated content and content written via artificial intelligence.

It should not be forgotten that Google has many years of experience in dealing with automation used in an attempt to deceive its search results, and efforts to combat spam, including through the SpamBrain algorithm (which by the way is based precisely on artificial intelligence), will continue regardless of how this spam is produced, Sullivan and Nelson assure.

However, it must also be acknowledged that not all use of automation is spam, and those exceptions include generation via artificial intelligence: “Automation has long been used to generate useful content, such as sports scores, weather forecasts and transcripts,” the guide says, and “artificial intelligence has the ability to fuel new levels of expression and creativity and serve as a critical tool to help people create great content for the web.”

This is the path Google is charting for publishers and SEO, then, in line with the way “we’ve always thought about empowering people through new technologies”: a responsible approach to using AI that leads to the creation of texts that ensure a high level of information quality to maintain the overall usefulness of content in Search.

Tips for creating AI-generated texts

Regardless of the content produced, those who aspire to achieve success in Google Search should always strive to produce original, high-quality, people-centric content by demonstrating EEAT qualities, which we know are the initials for Experience, Expertise, Authoritativeness, and Trustworthiness.

To keep up with what Google’s systems are trying to reward, then, the advice is to evaluate our content in light of these “parameters,” whether we decide to write manually or in the case of AI-assisted generation.

Basically, Google is telling us how to use AI, which can be a useful tool to support content creation, but cannot replace the human work of evaluating information, which is (or should be) the guarantee that content is useful and people-centered.

On closer inspection, this is the same line with which we encourage the use of our tool in the Editorial Assistant, as has been said since the launch of this feature and as is also stated in the disclaimer in the suite.

Google’s FAQ on AI content and Search

To further clarify its position about AI content applied to Search and SEO, Google has also published a series of FAQs, which make us better understand some aspects and provide more precise assistance and guidance.

      1. Is AI content against Google Search guidelines?

The appropriate use of artificial intelligence or automation is not contrary to Google’s guidelines. Appropriate means that Artificial Intelligence is not used to generate content with the main goal of manipulating Search rankings, because this is contrary to the search engine’s anti-spam rules.

      2. Why does Google Search not ban AI content?

Automation has long been used in publishing to create useful content, and artificial intelligence can lend support and generate useful content in new and exciting ways.

      3. How will Google Search prevent poor quality AI content from dominating in search results?

For Google Search, poor-quality content is not a new challenge to face: algorithms have been countering poor-quality content created by both humans and automation for years, and several systems (which continue to be regularly improved) are in place to determine the usefulness of content, while other systems work to elevate original news stories.

       4. What will Google do with AI content that potentially propagates misinformation or contradicts the consensus on important topics?

These problems exist in both human-generated and AI-generated content. Regardless of the content produced, Google’s systems seek to surface high-quality information from reliable sources, not information that contradicts an established consensus on important topics. On topics where the quality of information is paramount, such as health, social, or financial topics (which fall under the popular YMYL topics), the systems place an even greater emphasis on signals of trustworthiness.

       5. How can Search determine whether artificial intelligence is being used to spam search results?

Google has a variety of systems, including SpamBrain, that analyze patterns and signals to identify spam content, however it is produced.

       6. Will AI content rank high in Search?

The use of artificial intelligence offers no special advantage to content. It is just content. If it is useful, helpful, original, and meets EEAT aspects, it might perform well in Search. If not, it may not get visibility.

       7. Should I use artificial intelligence to generate content?

If you think artificial intelligence can be an essential way to help you produce useful and original content, it might be worth considering. If you see artificial intelligence as a cheap and easy way to distort search engine rankings, then no.

      8. Should I add author signatures to all content?

In general, you should consider showing accurate author signatures when readers reasonably expect it, such as for any content where someone might ask, “Who wrote this?” As a reminder, publishers appearing in Google News should use signatures and author information.

      9. Do I need to add statements about the use of AI or automation to my content?

Disclosures about the use of artificial intelligence or automation are useful for content where a reader might ask, “How was it created?” Therefore, the advice is to consider adding these details when it might be reasonably expected.

      10. Can I designate AI as the author of the content?

Giving AI an author signature is probably not the best way to follow Google’s recommendation to make it clear to readers when AI is part of the content creation process.

Barry Schwartz on Search Engine Land added another question, posed personally to Google and Danny Sullivan:

      11. Can artificial intelligence write content that demonstrates expertise?

Google sidestepped the question, reminding that “not all content has to show experience, expertise, authority and trust,” as specified in the guidelines to quality raters, and indeed it is not always the case that every piece of content achieves all the criteria (especially when it comes to distinguishing Expertise and Experience, i.e., Expertise acquired over time including through studies and training and, on the other hand, Direct, first-hand Experience).

In any case, Google reiterated and clarified that Trust is the most important part of EEAT, and that the other aspects contribute to Trust, but the content does not necessarily have to demonstrate all of them.

New also debuting AI features applied to Search

At the same time as the announcements about the evaluation of AI-generated content for ranking purposes on Search, Google also announced the launch of new, also AI-based features to improve Google Lens and multisearch, among other things, continuing in the wake of what was also announced at Google I/O 2022, for example.

With the new updates, users can search using both images and text simultaneously, as well as search directly from their mobile screens with Google Lens.

As Elizabeth Reid, VP of Search, puts it, the goal is to make “information exploration more visual, natural, and intuitive. “Artificial intelligence has played a crucial role in Google’s search technology since the early days, improving algorithms’ ability to understand language, and thanks to continued investment on this front Google is now able “to understand information in its many forms, from understanding language to understanding images, understanding video, and even understanding the real world.”

The latest Lens updates

A first way Google uses artificial intelligence is through its Lens feature, which has become increasingly popular with more than 10 billion monthly searches. With the latest update, available globally on Android devices in the coming months, users can use Lens to search for information directly from the screens of their mobile devices by framing an image with the camera and launching the search in the websites and apps they use (such as messaging and video apps) without having to leave the app or experience.


The refinement of Multisearch

The Multisearch feature (also discussed at Search On 22) has been talked about for some time, and finally this feature-which allows users to search with an image and text at the same time, opening up completely new ways for users to express themselves – is available globally on mobile devices, in all languages and countries where Lens is available.

For example, the article says, we might search for “modern living room ideas” and see a coffee table we like but would prefer in a different shape, such as a rectangle instead of a circle. Thanks to multisearch, just add the text “rectangle” to launch a search find the style we want.

In addition, Google is refining multisearch by adding the ability to search locally: if you take a picture and add “near me,” you can find what you need among geolocated results.

Google, AI and SEO: no revolution, just evolution

Google is creating more natural and visual search experiences, but it has “only scratched the surface,” Reid concludes, so “in the future, with the help of AI, the possibilities will be endless.”

Back to the Search and SEO world, here, too, Artificial Intelligence could find wide application, but the path laid out by Google (and previous experiences of countering automated techniques for ranking manipulation, such as with content farms, recalled from this article) lead us to assume that, ultimately, there will be no particular “revolution,” but rather an evolution through new technologies.

As mentioned, Google does not intend (nor can it) prohibit the presence of AI content among the search results, but it will also apply the criteria, signals and directions that apply to all others to the pages thus created.

Therefore, the reference must always be the same: we must write quality content for people, and that is how we will be “rewarded”. How and who writes that content do not matter much, because Google does not particularly distinguish between human or machine in terms of creator, if (and we emphasize if) if the content is written to help the people who read it and offers them useful information.

To summarize, Google does not give privileges (nor, at least officially, disadvantages) to AI content that hits the goals behind the concept of “quality.” Thus, if we use AI to find creative ways to add additional context and information that can help our readers, we can improve our pages and get positive hits in SERPs. Conversely, if we only use artificial intelligence to quickly, cheaply and efficiently produce a ton of content just to rank pages on Google, then we cannot hope too much for positive outcomes.

Google is (should be) good enough to detect content, whether produced by machines or humans, that is designed only to manipulate search, and acts accordingly.

At the end of the day, then, there is no revolution for SEO, as we said: the Google search algorithms will continue to analyze, evaluate, and rank content with the same rules, regardless of who actually creates it, be it Artificial Intelligence or human counterparts.