Google Multitask Unified Model (MUM)

The Multitask Unified Model is an AI-powered algorithm designed to use machine learning to deliver better search results to complex queries. Regarded as a milestone in AI, it is one of the biggest updates Google has launched to date. 

A 1,000 times more powerful than BERT, Google’s Natural Language Processing update of 2019, MUM enables Google to understand language in a more semantic, contextual and human way. In fact, it not only understands language, but generates it. It can piece together different bits of information in order to give precise answers to complex search queries.

Not only will Google be able to understand what people want to know better than ever, it will also be able to process the information on the web in greater depth. This means that users will be able to find relevant information, which may not necessarily mention their search term, but answers their query.

MUM is multimodal, meaning it understands information across text and images. And it will expand to more modalities like video and audio in future. Being trained across 75 different languages, it is designed to develop a more comprehensive understanding of information than previous models and deliver more accurate search results.

To illustrate how powerful MUM is, let’s say you’re looking for information on an obscure restaurant in Paris. Google MUM will automatically be able to translate a local French website with content about the restaurant and provide the information you’re looking for in English.

And it also works in reverse. For someone who does not speak English and who needs information on visiting a city in the US, MUM will translate content into their native language.

This is a complete revolution in how information is used across the internet and the level of access cultures will have to worldwide content. 

One of the biggest questions on everyone’s minds when it comes to SEO is: when is Google going to roll out MUM?

Google is still in the very early stages of developing and implementing MUM. As every improvement in Google Search undergoes rigorous evaluation and testing, it can be expected to be months or even years before MUM becomes fully realised.

Regarding this topic, Google said that:

“Just as we’ve carefully tested the many applications of BERT launched since 2019, MUM will undergo the same process as we apply these models in Search. Specifically, we’ll look for patterns that may indicate bias in machine learning to avoid introducing bias into our systems. We’ll also apply learnings from our latest research on how to reduce the carbon footprint of training systems like MUM, to make sure Search keeps running as efficiently as possible.”

This means that SEO strategists still have ample time to make adjustments and continue to plan for the future of SEO, no matter what it is going to look like.

While there will likely be big changes in SEO, consistently publishing informative and engaging content that meets search intent is expected to remain important.

Here are a Few Possibilities in a Post-MUM Era
Keywords might become less important – MUM is extremely intuitive, making it much like a personal assistant in web search that engages in a conversation with the user. As MUM will be able to pull key pieces of information from a wide range of media, it is likely that search results will be less dependent on keywords.
Written content no longer of primary importance – while written content like blog posts will always be significant, we may see more video, image and audio content featuring in search results. 
More competition in organic search – SEO may become more challenging in the sense that with MUM, there are no language or regional barriers. This means that your content is competing against those halfway around the world. 
Local SEO will become more important – with MUM being able to process content in 75 different languages, it’s likely that it’ll become necessary to focus more on the regional aspect of content in order to link it to a specific city or area.

To get a complete, well-rounded answer to nuanced or complex questions, you currently have to do an average of eight search queries.  

With MUM, Google aims to reduce the number of search queries by giving users more sophisticated, nuanced answers. It will do this in four ways:

MUM is language agnostic – the best answer may not always be in the searcher’s own language. To get around this, MUM searches for answers from 75 different languages.
Processing information across different media types – MUM goes beyond the written word and places greater importance on image, video and audio content to deliver relevant search results.
Providing predictive answers – it takes into account the full, multi-layered search intent behind a user’s query in order to deliver related, tailored answers and content, without the need to ask a series of subsequent questions.
Providing a unique, tailored answer by combining multiple sources – MUM is capable of producing unique prose that summarises information from multiple sources containing the most important, relevant points.