Here is MUM, the new algorithm that promises to change the interaction with the world’s most used search engine, namely Google Search
MUM first of all stands for Multitask Unified Model, a project that Google made official at the Google I/O event last May. It is a new algorithm that the company wants to “unleash” to ensure a better understanding of complex texts. In fact, Google has developed a model inspired by the nerve connections of the human brain, and thanks to a large number of extra nodes, MUM can be 1,000 times more powerful than the current system called BERT. From the moment MUM was born – and it can be assumed that this happened several weeks or several months before the announcement at the last Google I/O – the company has launched a series of pilot projects to test its capabilities, efficiency and future potential in online searches. And it’s precisely these that are prompting the more curious to try and understand how internet search can change thanks to MUM.
How MUM will change internet search in the short term
MUM will presumably run Google’s daily business for several years from the moment it is entrusted with the keys to the largest “library” on the planet, Google Search. So its adoption suggests improvements in the short, medium and long term.
In the short term it will change the efficiency of Google Search, which will be able to handle queries in less time, probably even imperceptibly for those who don’t pay too much attention to these kinds of details. MUM has been tested to optimize searches for vaccines, which present Google’s algorithm with a tough challenge because of the different names by which serums are known in different parts of the world.
MUM has identified more than 800 variants of vaccine names in more than 50 languages in a matter of seconds.
How MUM will change Internet search in the medium term
MUM has the potential to create entirely new ways to search and explore information. “Just like BERT,” said Pandu Nayak, vice president of Google Search, “MUM can understand language. Thanks to a series of decoders, however, it can also generate text, but that’s not all: because it’s multimodal, it can also understand images.”
In other words, Google’s new Search algorithm should be able to understand text and correlate it to images, which it would be able to “decode” in a way similar to how it does with text or even how a human being would do it.
How MUM will change Internet search in the long run
In the long run, Google expects a lot from MUM. Such as a deep understanding of information and requests from web visitors, and answers accordingly closer or more relevant to what is in the expectations of those who use the search engine.
MUM will aim to make it more “human” or intelligent if you prefer, dropping the (current) need to summarize and simplify as much as possible the requests to the web giant.