Are your SEO strategies failing to deliver the expected outcomes, and you're struggling to determine why?

Traditional SEO techniques are losing effectiveness daily. While you concentrate on keywords and backlinks, Google's AI continues to evolve swiftly, fundamentally altering how search rankings are determined.

This transformation occurs quietly, making it increasingly challenging to comprehend why your content isn't performing optimally.

Grasping how Google's AI systems function is crucial for adjusting your SEO strategy. This article examines the progression of Google's AI—RankBrain, neural matching, BERT, and MUM—and elucidates how these developments are reshaping search.

By comprehending these ideas, you'll be better prepared to produce content that aligns with Google's AI-driven methodology, enhancing your likelihood of achieving higher search rankings.

Google's AI systems

Since approximately 2015, Google has employed some form of AI to identify, evaluate, and arrange URLs, initiating with its first AI system known as RankBrain.

Three years later, Ben Gomes, Google's Senior Vice President of Learning and Education and former Head of Search, referred to AI as the "next chapter of Search."

Gomes stated that AI would enable Google to offer a superior user experience, not confined solely to the query. He mentioned AI would bring about "three fundamental shifts" in how search operates:

• From answers to journeys: “To assist you in resuming tasks where you left off and exploring new interests and hobbies, we’re introducing new features in Search that cater to ongoing information requirements.”
• From queries to providing a queryless method of accessing information: “We can present pertinent information tied to your interests, even when you don’t have a specific query in mind.”
• From text to a more visual means of discovering information: “We’re incorporating more visual content into Search and entirely revamping Google Images to facilitate easier access to information.”

This change commenced with RankBrain.

RankBrain (2015)

The RankBrain system marked the initial phase to aid the search engine in "understanding how words connect to concepts."

Grasping the relationship a word has with a concept is an intelligent endeavor, and Google's first step toward comprehending content akin to a human.

For instance, if you search “What’s the color of the sky?” the AI could discern what “sky” is and that it has a perceived color. Thus, Google could yield a result devoid of the exact words yet still answering the query.

A few years later, Google made further strides in linking words to concepts via neural matching.

Neural Matching (2018)

This system/sub-system was devised to assist Google in comprehending how “queries relate to pages” for concepts that are more intricate to decipher.

Suppose you search “tie my laces,” which could signify various meanings. With neural matching, Google could deduce that “l(fā)aces” refers to shoe laces and provide results on methods to tie them.

BERT (2019)

BERT stands for Bidirectional Encoder Representations from Transformers and was deemed a “breakthrough.”

Think of BERT as the progression of RankBrain and neural matching, enabling Google now to grasp how multiple words in a sentence relate to multiple words on the page and the concepts underlying them.

BERT appears to be pivotal for entity recognition. This capability aids Google in identifying a brand name, recognizing who a person is, and perhaps even determining their expertise in a particular subject.

This kind of AI model makes generative AI and AI Overviews feasible. Google has been utilizing it since 2019.

• Related to BERT is a “deep learning system” named DeepRank. As we learned from Panda Nayuk’s testimony during the DOJ trial, essentially DeepRank is BERT when BERT is used for ranking.
• DeepRank also supplanted much of RankBrain.

MUM (2021)

Google asserts that the Multitask Unified Model (MUM) is “1,000 times more potent than BERT.”

If BERT comprehends language, then MUM generates it. Additionally, it can grasp both text and images and potentially video by now.

Pandu Nayak, Google’s Chief Scientist, Search and former VP of Search, explained MUM thusly:

“Consider the query about hiking Mt. Fuji: MUM could recognize you're comparing two mountains, so elevation and trail data might be pertinent. It could also interpret that, within the context of hiking, to ‘prepare’ might involve fitness training as well as acquiring the proper gear.

Since MUM can uncover insights based on its extensive knowledge of the world, it could emphasize that while both mountains share similar elevations, fall is the rainy season on Mt. Fuji, so you might require a waterproof jacket.”

Nevertheless, MUM’s application to enhance search results regarding COVID-19 vaccine information underscores how powerful this system is.

Nayak noted that MUM aids in distinguishing between various vaccine brand names and furnishes the “most current trustworthy information about the vaccine.”

MUM highlights that Google can refine search results more expeditiously than previously.

Leveraging AI for SEO: What’s feasible?

What you can accomplish with generative AI, Google can execute with the AI in its ranking mechanism. Ponder that notion.

ChatGPT might possess an IQ of up to 155, so it's reasonable to presume that Google’s AI can scrutinize sources like a human to a certain extent.

A human assessing the quality and relevance of a page to their intent might pose these inquiries:

• Are you an experienced authority in the subject you're discussing?
• Are other seasoned experts discussing you and your expertise?
• Do you have a reputation for spamming Google to ascend rankings?
• How does what you say about a topic align with other experts in the field?
• Is this the finest product for what I'm searching for?

Yet, recall that Gomes stated AI will transition “From answers to journeys.” This is significant, implying that Google can trace how you and your audience interact with or generate content about your brand or internal experts.

Consequently, Google could address far more pertinent questions:

• Do individuals profit from your product or service?
• Is one website/company connected to another or distinct, with customers utilizing both?
• Are customers sharing details about your product and then seeking it on Google?

It’s time to cease contemplating SEO in terms of ranking indicators and focus on how humans seek information and why.

Contributing authors are invited to craft content for Search Engine Land and are selected for their expertise and contribution to the search community. Our contributors operate under the supervision of the editorial team, and submissions are verified for quality and relevance to our readers. The viewpoints expressed are their own.

The above is the detailed content of How Google Search uses AI. For more information, please follow other related articles on the PHP Chinese website!