How BERT Will Change the Way You Search
Google has been working on changing the way the internet responds to searches. Where Google once preferences preferred mobile-first technology, it has now switched partners and opted for AI-first technology. Artificial intelligence is making internet searches quicker and more accurate in yielding desired results.
How many times a day do you turn to Google to search for something? At any given time, it seems like we want to find out who someone is or what’s going on. The top three most popular keywords on Google, however, are weather (24,440,00 monthly searches), maps (20,400,00 monthly searches), and translate (18,310,000 monthly searches).
Google isn’t the only search engine available to students. There are plenty of alternatives for searching. Google, however, seems to be one of the best for yielding desired results.
Your internet searches are making Google one smart cookie, thanks to artificial intelligence.
For quite some time, algorithms have quietly worked their way through search engines, analyzing, and ranking the keywords. This newer search-ranking system is the Bidirectional Encoder Representations from Transformers (BERT). Bert arrived in the search room in October 2019.
BERT is the artificial intelligence algorithm designed to understand subtleties in language. The program’s algorithms can discriminate between the use of prepositions like “to” and correctly determine relationships between words and phrases. It reads nuances.
Possibly the most advanced AI program on the web today, BERT will change how our students use search engines.
Before Google launched BERT on the web, students learned a few simple tricks for searching online. Students searched exact phrases placed inside quotation marks. They also limited their searches to specific sites. The idea was to think like the computer, but sometimes those searches went awry. Misspelled words and inaccurate terms slowed the search process.
Already, though, Google was working to meet the needs of those searching online. The search engine assisted with definitions and suggested alternative combinations of keywords in an effort to return relevant results. Sometimes, it delivered. Other times, it didn’t.
Google responded with sites it thought you most wanted. These sites might have been based on previous searches. Other times, they were based on your location. Already, Google was trying to infer what you were looking for.
Now BERT is trying to think the way you would. It’s using inference to guess what you’re trying to find.
With BERT leading the way, students can use natural language for their searches and get good results that are more targeted and relevant. Students can type the way they talk, and their queries will turn up results that more accurately match what your students were looking for. The result is an impressive 10% change in search relevance.
You may have already noticed a change in the results you received for today’s searches. In addition to improving the relevance of its search results, Google’s BERT is beginning to feature snippets.
As our skills in using Google search improve, so will BERT’s ability to interpret them.