Google aims to answer more complex search queries through a new technology called the “Unified Multitasking Model” (“MUM” for short).
Google revealed at the I / O conference on Tuesday that it is exploring how to use MUM to better understand language and provide helpful answers.
MUM has yet to be released, but Google has conducted rare research on how it was developed.
Also read: conversion.ai review 2021
Google MUM Purpose
The goal of Google’s MUM is to help users complete tasks with less search volume.
On average, Google found that people need eight searches to complete complex tasks.
Google envisions that MUM can answer complex questions such as: “I have climbed Mount Adams and now I want to go for a hike. Next fall, Fuji, what should I do differently?” According to the current status quo, users do not have different information to answer the question, such as the altitude of each mountain, the average temperature in autumn, the difficulty of hiking trails, the use of appropriate equipment, etc.
Google compared MUM’s ability to convey answers to the same questions that hiking experts would answer.
MUM can understand that the user is comparing two mountain peaks, indicating that the altitude and trail information may be related.
Then it can be understood that “preparation” can include activities such as physical training and finding the right equipment.
In addition, MUM can emerge based on a deep knowledge of the world, which means that it can signal information such as that autumn is the rainy season in the mountains. Fuji, so users may need a waterproof jacket.
MUM can display useful subtopics for further exploration, such as the best equipment or the best training exercises.
The search results will guide users to useful articles, videos, and images on the web.
Also read: Content Marketing Strategy
Google MUM Technology
Like BERT, MUM is based on the Transformer architecture, but it is said to be 1000 times more powerful and can unlock information in various ways and in new ways.
Google says that MUM can understand information in different languages and formats to give users the most comprehensive results.
Google has started an internal MUM pilot and promised to apply this advanced AI to search results responsibly.
MUM will undergo the same rigorous testing as every Google search update before going online.
The company said it will pay particular attention to patterns that indicate biases in machine learning.
In the coming months and years, look for MUM-compatible features and enhancements to Google products.