A more “smart” Google will help you solve “personal crisis”

When there are questions and information they want to know, people are accustomed to using search engines to find the answers they need. In order to find this information for us faster, more accurately and better, search engines also need to continuously optimize their algorithms. For example, Google will integrate the latest machine learning model MUM, which will be launched in 2021, into the search engine.

▲ Picture from: Google

MUM based on Transformer architecture, the full name of Multitask Unified Model (Multitask Unified Model), cleverly uses 75 different languages ​​for multi-task training. Compared with previous models, it can understand information more comprehensively .

▲ Picture from: Google

MUM models can do more than just literal understanding and judgment. After learning structure and multiple languages, it can better understand complex languages. In other words, it can read your "subtext".

Google plans to use artificial intelligence in the future to improve the way it handles personal crisis searches in other countries. If you search Google for "the most common way to complete suicide" or "suicide hot spots", etc., the early system may understand information search , which now shows a message box offering help, which could be the phone number or website of a mental health charity.

▲ Picture from: Unsplash

Anne Merritt, Google's product manager for health and information quality, said the integration of MUM into the search engine enables the discovery of those queries related to individual circumstances that earlier search tools could not. Google also wants to do more to guide people to the information they need when it comes to suicide, sexual assault, and domestic abuse.

Of course, MUM can do more than help deal with "personal crises." Familiar with multiple languages, it can break through the barriers of different languages, and even use pictures and other methods to "guess" the information you need, instead of just sticking to words.

▲ Picture from: Google

For example, you want to query information related to mountain climbing. You can type a long paragraph and tell it that I have already climbed this mountain, and now I want to climb another mountain, what should I do differently?

At this time, MUM will start to "think". You may be comparing these two mountains, so you may need information such as altitude and path; you are inquiring about how to prepare, and you may need suitable equipment.

After "thinking", the results it shows may tell you that the mountain you want to climb is in the rainy season, you may need waterproof clothing, or it may tell you that there may be equipment you need in these articles or videos.

▲ Picture from: Google

If you take a picture of hiking boots, and then use your voice to ask a search engine: "Can I use it to climb Mount Fuji?" After MUM's "thinking", it may think you are looking for mountaineering equipment, and then display a Recommended equipment list.

▲ Picture from: Google

In addition to leveraging MUM, Google is also using the AI ​​language model BERT to better identify searches for explicit content such as pornography. By leveraging BERT, Google said, those "shocking results" were reduced by 30% year over year.

However, using artificial intelligence to improve search engines also has certain limitations. For example, machine learning language models may also contain biased or incorrect information in search results. Although there is still room for improvement, MUM that will learn does make Google's search "smarter", and for most users, this change is needed.

#Welcome to pay attention to the official WeChat account of Aifaner: Aifaner (WeChat: ifanr), more exciting content will be brought to you as soon as possible.

Love Faner | Original link · View comments · Sina Weibo