To celebrate 10 years of Creator Weekly, I’m sharing tech highlights from 2015 that still resonate 10 years later. This update was for the week of October 31, 2015.
Image by Google Deep Mind (via Pexels), "An artist’s illustration of artificial intelligence (AI). This image represents how machine learning is inspired by neuroscience and the human brain."
This was Google’s first AI-powered system for Search ranking, designed
to help the system understand how words are related to real-world
concepts.
What does that mean? An example Google used is a search for the “title of the consumer at the highest level of a food chain”.
Previously search was just looking at the words in a query in isolation.
As some put it, RankBrain looks at “things”, rather than “strings”.
In 2016 Google told Wired magazine that “... RankBrain is “involved in every query,” and affects the actual rankings “probably not in every query but in a lot of queries.” And “Of the hundreds of “signals” Google search uses when it calculates its rankings … RankBrain is now rated as the third most useful.”
Since 2015 Google has added a number of other AI systems in Search.
Of course Search Engine Optimizers (SEOs) want to know how to optimize for RankBrain and BERT and AI Overviews. And the answer is always the same from Google: make "helpful, reliable, people-first content".
Jack Clark @ Bloomberg, 26 October 2015,
Google turning its lucrative web searches over to AI machines
Steven Levy @ Wired, 22 June 2016, How Google is Remaking Itself as a “Machine Learning First” Company
10 years ago Google deployed Rank Brain, its first AI-powered ranking system
for Search.
RankBrain Looks for Things, not Strings
Image by Google Deep Mind (via Pexels), "An artist’s illustration of artificial intelligence (AI). This image represents how machine learning is inspired by neuroscience and the human brain."
On October 26, 2015
Google announced
that "a very large fraction" of search queries over the previous few months
had been interpreted by a machine learning system called RankBrain.
What does that mean? An example Google used is a search for the “title of the consumer at the highest level of a food chain”.
The system learned that “food chain” has to do with animals, so the searcher
isn’t looking for human consumers. By understanding and matching the words to
their related concepts, RankBrain understands that the search is looking for
what’s usually called an “apex predator.”
Previously search was just looking at the words in a query in isolation.
As some put it, RankBrain looks at “things”, rather than “strings”.
In 2016 Google told Wired magazine that “... RankBrain is “involved in every query,” and affects the actual rankings “probably not in every query but in a lot of queries.” And “Of the hundreds of “signals” Google search uses when it calculates its rankings … RankBrain is now rated as the third most useful.”
The View from 2025
- Neural Matching was added in 2018, to help understand “fuzzier” concepts in queries and match them to pages.
- In 2019 Google added a “neural network-based technique for natural language processing pre-training” called Bidirectional Encoder Representations from Transformers (which Google and everyone else calls BERT).
- In 2022 they added the Multitask Unified Model (MUM) which is capable of both understanding and generating language. That’s used for things like improving featured snippets.
Of course Search Engine Optimizers (SEOs) want to know how to optimize for RankBrain and BERT and AI Overviews. And the answer is always the same from Google: make "helpful, reliable, people-first content".
References
Google Search Central:
A Guide to Google Search Ranking Systems
Steven Levy @ Wired, 22 June 2016, How Google is Remaking Itself as a “Machine Learning First” Company
Google Blog, 3 February 2022,
How AI powers great search results
Comments
Post a Comment
Spam and personal attacks are not allowed. Any comment may be removed at my own discretion ~ Peggy