How Google uses artificial intelligence in Google Search


As Google continues to leverage more artificial intelligence and machine learning in Google Search, one wonders how AI and machine learning are helping Google Search perform its day-to-day tasks. Since 2015, when Google introduced its first AI in search named RankBrain, Google has continued to deploy AI systems to better understand language and thus improve the search results that Google presents to its searches.

Several months ago, we sent Google a number of questions about how Google uses its AI in search, including RankBrain, Neural Matching, BERT, and Google’s latest breakthrough in AI – MUM . We got a better understanding of when Google uses AI, what AI does what in Google Search, how these different AI algorithms can work together, how they’ve changed over the years, and what search marketers need , if applicable. learn how Google uses AI in search.

We spoke with Danny Sullivan, the Public Liaison for Google Search, to help us find answers to many of these questions. In short, RankBrain, neural matching, and BERT are used in Google’s ranking system for many, if not most, queries and seek to understand the language of the query and the content it ranks. However, MUM is currently not used for ranking purposes, it is currently only used for naming COVID vaccines and feeds related topics in video results.

It starts with writing content for humans

You hear it all the time from Google reps and many SEOs: write content for humans. Back in the days of SEO, when the algorithms were maybe simpler, you had lots of SEOs creating content for every search engine (back then there were dozens of different search engines). Now there’s mainly Google, with a bit of Bing and a few ruffles of DuckDuckGo – but the algorithms are much more complex and with machine learning and AI, the algorithms understand language more like a human would understand language.

So the advice Google gave is to write for humans, and that you can’t optimize your site for BERT or any AI. If you write content that humans understand, AI algorithms and search engines will understand it too. In short, this article is not intended to try to give you SEO tips on how to optimize your sites for a specific AI, but rather to communicate how Google uses AI in Google Search.

Overview of the AI ​​used in Google Search

RankBrain. It starts with RankBrain, Google’s first attempt to use AI in search was in 2015. Google told us that RankBrain helps Google understand how words relate to concepts and can take a broad match and better define how this query relates to real-world concepts. While it was launched in 2015 and was used in 15% of queries, Google said that it is now, in 2022, widely used in many queries and in all languages ​​and regions. RankBrain specifically helps Google rank search results and is part of the ranking algorithm.

  • Launch year: 2015
  • Used for filing: Yes
  • Examines the query and content language
  • Works for all languages
  • Very commonly used for many queries

Here is an example provided by Google of how RankBrain is used, if you search for “what is the title of the consumer at the highest level of a food chain”, Google’s systems learn by seeing these words on different pages that the concept of a food chain may affect animals, not human consumers. By understanding and matching these words to their associated concepts, RankBrain helps Google understand that you are looking for what is commonly referred to as an “apex predator”.

Neural pairing. Neural matching was the next AI released by Google for research, it was released in 2018 then expanded to local search results in 2019. In fact, we have an article explaining the differences between RankBrain and Neural Matching here. Google told us that neural matching helps Google understand how queries relate to pages by looking at the entire query or page content and understanding it in the context of that page or query. Today, neural matching is used in many, if not most, queries for all languages, in all regions, in most search sectors. Neural matching specifically helps Google rank search results and is part of the ranking algorithm.

  • Launch year: 2018
  • Used for filing: Yes
  • Examines the query and content language
  • Works for all languages
  • Very commonly used for many queries

Here is an example provided by Google of how neural matching is used, if you search for “insights how to manage a green”, for example. Google said “if a friend asked you this, you’d probably be confused”. “But with neural correspondence, we are able to make sense of this interrogative research. By looking at the broader representations of the concepts in the query – management, leadership, personality and more – Neural Match can decipher that this researcher is looking for management advice based on a popular color-based personality guide,” we said. Google.

BERT. BERT, Bidirectional Encoder Representations of Transformers, came in 2019, it is a neural network based technique for pre-training in natural language processing. Google told us that BERT helps Google understand how combinations of words express different meanings and intent, including looking at the sequence of words on a page, so even seemingly unimportant words in your queries are taken into account. When BERT was launched, it was used in 10% of all English queries, but spread to other languages ​​and was used in almost all English queries from the start. Today it is used in most queries and is supported in all languages. BERT specifically helps Google rank search results and is part of the ranking algorithm.

  • Launch year: 2019
  • Used for filing: Yes
  • Examines the query and content language
  • Works for all languages, but Google said BERT “plays a vital role in almost every query in English”
  • Very commonly used for many queries

Here is an example provided by Google of how BERT is used, if you search “if you search “can you get drugs for someone pharmacy”, BERT helps us understand that you are trying to figure out if you can take drugs drugs for someone else. Before BERT, we took this short preposition for granted, mostly showing results about how to fill a prescription,” Google told us.

MOM. MUM, Multitask Unified Model, is Google’s newest search AI. MUM was introduced in 2021, then expanded again in late 2021 for more applications, with many promising uses in the future. Google told us that MUM helps Google not only understand languages, but also generate languages, so it can be used to understand variations of new terms and languages. MUM is not currently used for any ranking purposes in Google Search, but supports all languages ​​and regions.

  • Launch year: 2021
  • Used for filing: No
  • No specific query or languages
  • Works for all languages ​​but Google is not used for ranking purposes today
  • Used for a limited number of purposes

Currently, MUM is being used to improve searches for COVID-19 vaccine information, and Google said it is “looking forward to providing more intuitive ways to search using a combination of text and images in Google.” Lens in the coming months”.

AI is used together in search, but can be specialized for search verticals

Google’s Danny Sullivan also explained that although these are individual AI-based algorithms, they often work together to help rank and understand the same query.

Google told us that all of these AI systems “are used to understand the language, including the query and potentially relevant results,” adding that “they are not designed to act in isolation to just analyze a query or a page”. Previously, it might have been assumed and understood that an AI system might have sought more to understand the query and not the page content, but that is not the case, at least not in 2022.

Google has also confirmed that in 2022 RankBrain, Neural Matching and BERT are used globally, in all languages ​​in which Google Search operates.

And when it comes to web search versus local search versus images, shopping, and other verticals, Google explained that RankBrain, neural matching, and BERT are used for web search. According to Google, other modes or verticals of Google search, such as images or the shopping mode, use separate and specialized artificial intelligence systems.

What about core updates and AI

As explained above, Google uses RankBrain, Neural Matching, and BERT in most queries you enter into Google Search, but Google also offers basic updates. The general Google updates that Google rolls out several times a year are often more noticed by site owners, publishers, and SEOs than when Google releases these big AI-powered systems.

But Google said it could all work together, with basic updates. Google said these three systems, RankBrain, Neural Matching, and BERT, are the biggest AI systems they have. But they have many AI systems in search and some in the core updates that Google is rolling out.

Google told us that they have other machine learning systems in Google Search. “RankBrain, Neural Matching, and BERT are just a few of our most powerful and important systems,” Google said. Google added, “there are other AI elements that may impact core updates that don’t affect these three specific AI systems.”

New to Search Engine Land

About the Author

Barry Schwartz, Search Engine Land editor and member of the SMX events programming team. He owns RustyBrick, a New York-based web consulting firm. He also runs Round table on search engines, a popular research blog on very advanced SEM topics. Barry’s personal blog is called Cartoon Barry and he can be followed on Twitter here.


About Author

Comments are closed.