Close

Will Google's MUM Redefine Search?

August 24, 2021

2 min 50 sec read
154 Facebook Twitter Whatsapp Facebook Twitter LinkedIn Pinterest WhatsApp Copy Link Your browser does not support automatic copying, please select and copy the link in the text box, then paste it where you need it.
When it comes to keeping it steady with algorithms, Google is like J. Lo, notoriously known to "play the field." Google changes its algorithm more times than the CDC's advice for the public on handling COVID.

Man With Google Logo Head Holding Jealous Sesame Street BERT's Hand Staring At Woman Named MUM
Google's thinking about dropping BERT, and they're texting testing your MUM, a new algorithm that may redefine how we search. We couldn't resist that last strikethrough.

So what's the Multitask United Model all about, and why is BERT getting the boot 🥾? Well, they're not really dropping BERT. They're upgrading it.

So in jibberish, BERT is a neural network-based technique for natural language processing pre-training and stands for "Bidirectional Encoder Representations from Transformers." But yeah, we'll just stick with BERT Google.

Ok, let's explain in English. BERT understands the relationships between words and their contextual meaning.

BERT's done a fantastic job in how we search, especially queries in a question format. You know, when we write them like this: "how to change your name and skip town?" or "why does my husband __?"

Anywho, BERT shook the world in 2018, but Google's new MUM algorithm could transform the way we search forever 🤯 — here's how.

With MUM activated on Google's search engine, searches are gonna look a lot less like game-winning answers on Jeopardy and more in tune to get things done for you. If you think about it, the name is starting to make sense.

MUM will be able to help you with your research to achieve simple and complex tasks like how to prepare yourself to climb Mt. Fuji.

Not only that, it'll recognize slang and shorthand language, so teenagers and stenographers finally have something in common.

Google recently launched their conversational AI language model called LaMDA, but they've taken it a step further with MUM. MUM knows trillions of words, and when continuously trained (fed data), it will get exponentially better and better at understanding the human language.

The model finds connections between words and patterns via different modalities. Don't worry. We had to google modalities too.

It's a fancy way of saying how you experience, learn, or express something and it may include 1 of the 5 senses.

MUM won't emit delicious smells out of your iPhone when you search "what's the best 🍕 near me?" But here's what it can do. It's trained to know 75 languages and aims to acquire the knowledge of the world.

It will fully understand and output text, images, video, and sound to help us find content and answers across different types of media.

Based on your previous searches and preferences, it can help you find new shows and movies to watch, books to read, new music and art to enjoy.

MUM could one day help parents find good schools for their children and even help others find a job. Or if you're looking to try out a new fitness routine, MUM is there to spot you — with a new fitness routine — thanks, mom!

Vice President of Search, Pandu Nayak, said that MUM would be able to help searchers with "fuzzy information needs."

Going back to our Mt.Fuji example. If you've wanted to know how to prepare yourself to climb it, you'd have to search on a lot of things like when's the best time to go and what to bring, what's the elevation, etc. with MUM, it's like talking to a hiking expert in one click of a search.

Google hasn't announced the timeline of their MUM algorithm yet, but they're already facing ethical backlashes from the tech community. They're saying MUM and large language models like it will learn and output queries with biases.

People have also criticized Google for planning to use MUM to keep search traffic to themselves, and Nayak denies that "It's not going to become a question answering [system]."

And "the content that's out there on the web is rich enough that giving short answers doesn't make sense," said Nayak.

Fewer searches needed to answer queries might lead to a limited amount of ad spaces for businesses. This could make paid search advertising much more expensive, but the more targeted the ad, maybe it'll work out?

Who knows how the algorithm will play out in the future.

We'll see about your MUM Google. Nailed it!

Categories: Google, Technology
154 Facebook Twitter Whatsapp Facebook Twitter LinkedIn Pinterest WhatsApp Copy Link Your browser does not support automatic copying, please select and copy the link in the text box, then paste it where you need it.