What Google MUM means for the future of search
Google’s 2021 I/O conference was an online-only event this year – but that didn’t scale down the size of the announcements. In his keynote speech, Google CEO Sundar Pichai revealed the introduction of a Multitask Unified Model, or MUM, which will mean Google can answer complex and nuanced questions, or even several questions at once, without multiple searches. This new milestone could have significant implications for the future of search and how brands serve customers online.
WHAT IS GOOGLE MUM?
A Google search query is often the very first step you take in planning a holiday, writing an essay, getting inspiration for a home renovation or simply learning something new. However, it typically takes up to eight searches – yes, eight! – to complete complex tasks. What if those eight individual searches could become just one, and still answer all your questions? That’s exactly where MUM comes in.
MUM uses advanced AI and is built on what is known as ‘transformer architecture’ – and no, we’re not talking about Optimus Prime! It’s a bit like the former Google search model, BERT, which launched in 2018, and used a neural network technique for natural language processing. BERT is set to evolve further; Google also announced LaMDA at I/O, another new technology that will make AI chatbot conversations more naturalistic. However, MUM will reportedly be 1,000 times more powerful than BERT, with 1,000 times more nodes, or decision points, and will have the capability to generate language. MUM will also be multimodal: for now, this means that Google will be able to process information in images as well as in text, but in the future this may extend to other capabilities including video and audio.
NEW UPDATES
You could be forgiven for wondering if there’s actually any more work left to be done to Google, given its sheer scale and dominance of the market. But the website has a constantly evolving roadmap of new updates, with around 1,000 changes to its search algorithms every year. While most of these are minor daily tweaks, often only affecting specific countries or regions, Google doesn’t publish when they happen or what they are for. It’s only a few times per year, when a broad core algorithm update is released, that we can gain more detail from Google about what these changes mean for advances in SEO, user experience and ranking performance.
Perhaps the most significant of these major updates for this year – and for some time now – has started to roll out this month on mobile devices. The Core Web Vitals (CWV) update will use user experience metrics as ranking signals (the individual factors Google uses to rank websites). Measuring and rating user experience metrics will encourage website owners to provide better experiences for users and help Google to display these sites more quickly and efficiently, so it’s a win-win all round. Google confirmed at I/O that they’ll be rolling out CWV across desktop following the release on mobile, so if the mobile-first implementation goes to plan, we can expect desktop to follow in September or early December 2021, given the site’s well-established development release cycles.
As yet, Google have not announced the timeframe of their development and release cycle for MUM. As it’s currently only in an internal pilot test stage, it may be a few years before MUM is fully released. When it is, it will be a major update and Google will notify the public in advance of any rollout. As Google’s Search Liaison Danny Sullivan has already all but confirmed, Google won’t be mum on MUM!
UNDERSTANDING COMPLEX SEARCH INTENT
At I/O, Google Search VP Pandu Nayak gave some context on the relatively simple multiple search queries that users perform today, compared to the complex queries that MUM will help to combine and solve more readily in the future. He gave the example of a user who’s hiked Mount Adams and is preparing to hike Mount Fuji next autumn. Google can help you with this today, but only after many thoughtfully considered searches. With each individual search, it could give you information such as the elevation of each mountain, the average temperature in the autumn months and the difficulty of hiking trails. Meanwhile, you could ask a hiking expert just one question – what should I do differently to prepare for my next hike? – and they’d give you a thoughtful answer that took into account all of these nuances.
MUM will be able to understand that you’re comparing two mountains, so elevation and trail information may be relevant. It will also understand that in context of hiking, ‘prepare’ could include both fitness training and finding the right gear. With these insights, MUM can highlight that both mountains are of roughly the same elevation, and that the autumn months are within the rainy season for Mount Fuji, so you might need a waterproof jacket. It can also provide helpful sub-topics for deeper exploration, like the top-rated gear or best training exercises, and direct you to helpful articles, videos and images from across the web.
Using Google right now, you can’t find information about Mount Fuji written in Japanese unless your search query is also in Japanese. But MUM will have the ability to learn from sources in other languages and use that information to provide the most relevant results in your chosen language. This means that future Google searches for visiting Mount Fuji might show results about where to enjoy the best views of the mountain, local highlights such as ‘onsen’ (hot springs) and popular souvenir shops – all information more commonly found when searching in Japanese.
THE FUTURE OF SEARCH
So, what will it mean for the future of search when Google MUM eventually rolls out? Well, quite simply, you’ll need to complete fewer searches to get things done, and you might combine a couple of unrelated search queries in the search bar. Eventually, you might even be able to take a photo of your hiking boots and ask ‘can I use these to hike Mount Fuji?’ and Google will let you know that your boots will work just fine, and then point you to a blog with a list of recommended gear.
Google is still in the early days of exploring the power and capability of MUM, so we don’t yet know if it will prove to be as transformative as they suggest it will. But it’s clear that MUM is a step towards a future where Google can understand all the different ways people naturally communicate and interpret information. What we define today as ‘search’ might ultimately evolve into something bigger as natural language processing and AI plays a bigger role. Search may begin to look more like research, as an injection of critical thinking turns Google into more of a knowledge presentation hub than the search engine we recognise today. Google won’t just know and organise where everything is on the web; it may become more like a personal research assistant with expertise across many specific topics.
As Google opens up the idea that search may involve input from many sources, such as microphones, cameras, TVs and wearable technology, search and search presentation will likely have to change. The introduction of MUM might simply mean better search results, but it may also mean a new type of search result that could completely revolutionise how we think and approach content. As less importance is placed on answering questions and solving problems, there may be a shift to a new, as yet unimaginable way of finding and consuming content. A future where we have to do less of the dirty work and get more of the enjoyment – for better or worse.
Want to be ready for the future of search? Get in touch.
