January 26, 2023

What does Google think about AI content?

In days gone by, Artificial Intelligence was the stuff of science fiction. These days, however, it's not about killer robots and sentient computers but more about how AI affects everyday life. From smart technology to Google searches, artificial intelligence plays its role. 


Recently, it has once again been a topic of discussion within SEO. As AI technology becomes more advanced, what does Google think about its place in content creation?


A brief history of Google's relationship with AI 


AI is an integral part of how Google ranks content. In 2015 Google introduced RankBrain, a machine-learning AI that could adjust results based on past information. In other words, it would learn about your search patterns and history and adjust results for your current query accordingly. And adding AI into the search algorithm didn't stop with RankBrain.


In 2019 they took their AI capabilities further with the introduction of BERT, a language-based machine-learning AI that contextualizes and interprets words in a search term to understand the intent of a query better. Finally, in 2021, Google added MUM and LaMDA to their AI toolkit. Short for Multitask Unified Model, MUM is another AI language model that can understand and generate language. This means it can take one search query and generate information that may be of interest. For example, if you search for "Flights to Paris" and later “Best Restaurants in Paris," MUM can interpret that you may be going on vacation and so generate other helpful information such as the weather, hotel information, and events. LaMDA, or Language Model for Application Dialogues, focuses on understanding the open-ended quality of conversational dialogue. In fact, you may even remember in June 2022, a Google engineer claimed LaMDA was sentient. 


It's safe to say that Google is a fan of artificial intelligence. That is if it applies to their algorithm and helps to improve search ranking for users. 

 

What they've not been a fan of is what they call auto-generated content. 


In the early 2010s, not long before Google's own application of AI, content farming was a real issue for the search engine. Essentially, content stuffed with keywords with little substance purely for the purpose of ranking in search results to boost the overall SEO of the site hosting it. Major companies that had made a fortune from content farming were affected by Google's Panda algorithm update, which targeted low-quality content and penalized sites that contained large volumes of it. 


Many would consider the next step up from content farming to be generative AI. Rather than having a real person create content designed to rank, you're simply automating the process. 


Google's Search Advocate, John Muller, has expressed in the past that while the quality of generated content has improved, it would still be considered spam by Google. That would mean that AI-generated content is against webmaster guidelines; however, it isn't easily detected by Google's filters and bots. And in mid-2022, they announced their "Helpful Content Update," which put even more emphasis on creating content "for people, not robots." In other words, genuine, valuable content that isn't created just for the purpose of ranking. 


That's been the story until late 2022, when the conversation around AI, its place in content creation, and Google's stance on the matter became the buzz of the SEO world once again.


Why is everyone talking about AI?


The use of Artificial Intelligence has long been a hotly debated topic in a variety of different fields. Within SEO and search, it has more recently been ignited by the rise in more advanced generative AI, such as ChatGPT. These new tools can produce long-form, more nuanced pieces of content than its fellow GPT-3 predecessors. In late 2022, this was the talk of the industry as people questioned its place in content creation and SEO.


In early 2023, the first major "scandal" of generated content in search came to light when it was discovered that Bankrate was creating large amounts of content using AI, some of which contained inaccuracies. When found, initially, it was met with little action from Google, citing that their issue was not with AI content but the usefulness of the content. Unfortunately, this inaction has led some bad actors to believe that this is Google's blessing to entirely AI-written content.  


So, the discussion continues. Is this the new content farming? What is the correct application of AI? Will it replace the need for content teams? How will this affect search and ranking in the future? 


What's Google's current stance on AI-generated content?


Following the news about Bankrate, Google's Public Search Liaison, Danny Sullivan, took to Twitter to clarify Google's stance:


"As said before when asked about AI, content created primarily for search engine rankings, however it is done, is against our guidance. If content is helpful & created for people first, that's not an issue.” 


Essentially, Google maintains its "people first" policy and guidelines around spam, whether or not content is AI-generated. It will rank if the content is focused on user experience, is helpful to users, and is high-quality by their rater standards. The reaction to this response was mixed, with some accusing Google of showing favoritism to a large corporation. 


However, a key point is that the Bankrate content was edited and reviewed by humans. This seems to be the current potential "good" application of AI - generate an article on a topic, followed by human editing and review. This may fall within Google's newly announced E-E-A-T guidelines, where content is also to be assessed by search quality raters on experience, as well as expertise, authority, and trustworthiness. 



This can allow for fast content generation, giving editors a starting point to add the nuance and experience only a human can provide. Where the balance lies between human effort and AI generation, however, is still up for debate. Should it be determined on a percentage basis? Or by what additional research and nuance have been added? Should it be purely based on the finished result and its overall usefulness to a reader? Should the AI origin of the article be flagged? These questions will undoubtedly arise and eventually be answered in the near future as AI content continues to develop.


—


The relationship between Google, AI, and content is changing as technology grows and people learn how best to apply it. In its current form, it certainly has its place. But the desire for high-quality content written with experience and expertise has an even more valuable place in creation and SEO. 


As the tech around AI develops and it becomes more accessible to content creators and marketing teams, we'll no doubt see innovation in this space. This may lead to more interesting, useful, and engaging content being produced at greater speeds. As that happens, we may see Google's policies and guidelines shift. Whatever the case, Google's relationship with AI is certainly one to watch in the months and years ahead.