There’s no getting away from it now: AI has firmly arrived, and it’s here to stay. While artificial intelligence has been around for decades at this stage, never has it been so prevalent and utilized. The refinement of this technology in recent years has seen it being used by companies, both big and small, to improve processes, speed up work, reduce
resources, and more.
One company that has benefited greatly from AI is the dominant search engine, Google. Their algorithms, which are used to retrieve data from their search index to supply users with the most relevant, accurate results for a query, have made giant leaps over the past decade after the introduction of AI technology.
We’ve spoken to Paul Morris, MD and leading Bristol SEO agency, Superb Digital to explain how AI is being used in search, the main components that make up Google’s AI approach,
and what the future holds with this technology for search engine results.
How AI transformed Google’s ability to search
Before the introduction of AI and during the early days of Google, their approach to search
was relatively rudimentary. That approach was to simply search for results that matched up
with the keywords used. For example, if you had a query for “chicken”, it would simply use
this word to match it up with relevant website pages.
While fine in theory, problems quickly cropped up with misspellings. Say you mistakenly
searched for “chikcen” instead; the results wouldn’t be favorable. Unless a page targeted
that particular misspelling – and in that situation, it would typically be created by a spammy site – you likely wouldn’t find what you were looking for. The result: you would have to redo the search and waste some precious seconds.
Now it might not seem like much, but Google wants to provide the best service to its users.
Even if it was a mistake on your part, they don’t want to slow you down or supply you with
irrelevant pages. With the introduction of advanced AI and machine learning, the search
giant has been able to improve its search systems to interpret human language much better.
As Google put it:
“With recent advancements in AI, we’re making bigger leaps forward in
improvements to Google than we’ve seen over the last decade, so it’s even easier for you to find just what you’re looking for.”
These search enhancements go beyond them just being able to understand misspelled
words. With AI at the forefront, Google can better understand passages – aka specific
searches that take the form of a sentence – to pinpoint the most relevant result and spotlight this at the top of the search results page. Similarly, AI has also helped Google provide more diverse content – and understand subtopics – whenever a broad search query takes place.
RankBrain was first launched by Google in 2015. At the time, it was a pioneering
development. The reason: this was the first time that Google’s search had incorporated a
deep learning system.
Aside from it being the first AI system implemented by Google, RankBrain was important due to its role in linking words to concepts. Instinctively, humans are well aware of how these words match up with concepts. For a computer, on the other hand, it is a serious challenge to comprehend the correlation – or at least, it used to be.
With RankBrain, Google suddenly gained the ability to find the information it couldn’t locate
previously, and this was because it could better perceive how certain words within a query
connected to real-world concepts.
As an example, say you were searching for an answer to the following question: “What’s the marketing process that is used to better improve a website’s visibility in search results?”. By analyzing different pages that contain those words, RankBrain is able to understand you are likely talking about online marketing rather than traditional marketing. Then by further matching the query words to related concepts, the system realizes you are searching for the answer: “search engine optimization (SEO)”.
As you might gather from its name, RankBrain is utilized by Google to rank search results
better and decide the best order for content. This means if you produce content in the hope
of landing on the first page of Google, you have to go beyond just keyword implementation
and add the concept – and context – behind said words.
It may have been introduced eight years ago, but RankBrain isn’t an obsolete piece of
technology. Even though modern systems have pushed AI to new levels, RankBrain remains an integral cog in the Google machine. In fact, it is still one of the main AI systems that makes the search engine what it is today.
BERT, which was introduced in 2019, further built on the work started by the aforementioned RankBrain. This AI technology further helped – and continues to help – with how Google can understand natural language, including how there are different intents and meanings behind combinations of words.
What does this mean? Well, rather than trying to match queries with content via individual
words, BERT can grasp how a complex idea is expressed by an entire word combination. As
a result, the system can comprehend a sequence of words, including how they all link to
each other, to provide more accurate answers to search queries. Rather than simply
focusing on the main keywords in a sentence, every word is included.
For instance, say you made the following search query on Google: “Can you buy a car for
someone else”. With the introduction of BERT into their search system, no word is left
behind with this query. It realizes that, with the preposition included, you are trying to find out how to purchase a car specifically for someone else – and not yourself. In the past, that
wasn’t the case, and you wouldn’t have been supplied with accurate results for this type of
Now, with BERT in place, Google has a better comprehension of small words and how these
can have a big impact on searches.
In 2021, Google debuted its latest AI system in its search engine structure. The Multitask
Unified Model, MUM for short, is described by the search giant as being one thousand times
more powerful than its predecessor BERT.
There are various reasons why this is the case. Firstly, it can both understand and generate
language. Secondly, it’s not just English where it does its work – it is trained to work across
75 languages. MUM can also complete a large number of tasks at once. Furthermore, it is
multimodal, which means it can understand information not just from text but also images
and other forms of content in the future.
MUM remains in the early stages of its development. To demonstrate this, it still isn’t used to improve search results and rank content – two tasks that are done by the RankBrain and
BERT systems. However, MUM is continually being incorporated into Google’s search, and it
will play an even greater role in search results in the future.
How will MUM and AI affect SEO?
The rollout of MUM has left certain marketers asking the question: ‘Is SEO on the way out?’
As noted by SEO agency Superb Digital, SEO will remain a valuable weapon in any online
marketer’s repertoire. Yet the introduction of MUM changes the way that SEO is done.
Rather than relying on keywords to prop up your content, getting to the top of search engine results pages will be dependent on the overall quality of what you produce.
Yes, content quality is already a key component of SEO. But this will only further grow in
importance in the coming years. This means content creators will have to adapt to MUM and the changes it continues to make. Some tips for doing that include:
- Produce content that answers the main questions being asked by your audience.
- Optimize structured data and make use of title tags and other elements.
- Diversify your content. As MUM is multimodal, you should add audio, video, and
image content alongside your text.
- Optimize your content for voice search. More and more people are using voice
search, which means adding conversational search terms to your content can make
a big SEO difference.
What the future holds
The significant improvements already made by AI are only the start. As this technology
continues to advance and evolve, it will play an even bigger role in how search works. While this is great for users as they will get more accurate, relevant results for their inquiries, it is something that will make businesses wary about their SEO efforts.
Reacting to search algorithm changes is nothing new for those in the digital marketing game. However, AI technology continues to modify the rulebook with sweeping changes, and marketers have to react to these changes – and fast. The good news: content remains king.
Due to this, if you are still producing high-quality content, the type your audience wants to
consume, you should be duly rewarded by Google. However, with the different SERP structure, greater competition due to more languages being added to the equation, and the move away from keyword reliance, there’s no guarantee of SEO success.