Google’s Newest AI Technology MUM Teased at 2021 I/O Conference

Google’s annual I/O conference began this past Tuesday with many interesting developments coming out. One of the most notable announcements was the introduction of a new technology called Multitask Unified Model (abbreviated to MUM). MUM builds on BERT, Google’s previous model, except it is 1000 times more powerful and has many more capabilities. The key difference between BERT and MUM is that MUM has the ability to multitask.

What is MUM?

In short, MUM is part of Google’s ongoing efforts to make online searches easier and more helpful for consumers. MUM is part of an initiative to allow users to make more complex searches than what AI can currently comprehend.

What can MUM do?

The main goal of MUM is to better understand human language. As AI becomes more intelligent and can better understand complex questions that we are asking of them, the more helpful it can be in delivering results. MUM can run numerous tasks at the same time, meaning that it can gather a deeper knowledge of the world and of the way that humans talk than ever before. The stand out features of MUM are that it both understands and generates language, is trained across 75 different languages, and is multimodal (meaning that it can understand multiple forms of information such as images, text, and in the future video and audio). This allows it to gain a deeper understanding of the world that will assist search engines in delivering solutions to more complex questions.

What does this mean?

Ultimately, MUM has the goal of allowing you to execute more complex searches. Instead of using multiple searches to find the information that you need, it can all be done in one. The example used by Google’s Senior VP Prabhakar Raghavan, who introduced the new technology at the conference, was the query: “I’ve hiked Mt. Adams and now want to hike Mt. Fuji next fall, what should I do differently to prepare?” While this would be too complex for current search engines to decipher, MUM is getting us closer to the reality of having this answered for us. You could have the similarities and differences between the two mountains highlighted for you, and could also be provided with advice on the right equipment to use on your hike. Raghavan also demonstrated the multi-modal aspect of MUM with the example of boots. Raghavan says: “Imagine taking a photo of your hiking boots and asking ‘Can I use these to hike Mt. Fuji?’” In this scenario, MUM would hypothetically be able to indicate whether the boots are ideal for hiking Mount Fuji and also point users towards a list of recommended equipment.

When can we expect MUM to launch?

If MUM works as demonstrated at I/O, it could revolutionise the way that search engines work and provide a more cohesive search experience for users. While there is currently no news on when MUM will be rolled out, Google has indicated that they have commenced internal pilots.

We are a friendly and reliable team that pays detailed attention to your projects and management of your brand. Our team at SEO Shark has a passion for all things online. We constantly innovate using the latest professional techniques and strategies.