How Tech companies are getting benefited using AI and ML in their Products/Services

In this article, I am going to discuss how MNCs are getting benefited from using AI/ML. Before going into the case study first I want to discuss something on AI and ML. So this gets into our first question

What is artificial intelligence (AI)?

AI refers to the ability of a computer or a computer-enabled robotic system to process information and produce outcomes in a manner similar to the thought process of humans in learning, decision making, and solving problems. So that it can function without the intervene of humans into the system.

‘Artificial intelligence is the science and engineering of making intelligent machines, especially intelligent computer programs.’ — John McCarthy, father of AI

John McCarthy

Now we know what is AI. Within AI we have these following domains…

source: PWC analytics

What is Machine Learning (ML)?

Machine learning is a part of AI, that deals in the training part of the system or machine.

Machine learning enables computer systems to learn from and interpret data without human input, refining the process through iterations to produce programs tailored to specific purposes.

Machine learning can also be used to run simulations, using predictive data models to discover patterns based on a variety of inputs.

Now we have defined ML, lets see what can we do using ML. There are different algorithms we can train our system, they are

The term machine learning was coined in 1959 by Arthur Samuel, an American IBMer, and pioneer in the field of computer gaming and artificial intelligence.

Supervised learning:

Let's take an example, Suppose your school teacher shows you some fruits and explains to you taking each fruit one at a time, Taking one fruit, and explains that “ This is an apple ”, taking another fruit and says “ This is mango” and so on...

Now that you know some fruits when we some fruits by the side of the road we can say which fruit is that. because we have an idea about it before we can say what fruit is it. This is called Supervised Learning.

A common example of this is image classification. Often, we want to build systems that will be able to describe a picture. To do this, we normally show a program thousands of examples of pictures, with labels that describe them. During this process, the program adjusts its internal parameters. Then, when we show it a new example of a photo with an unknown description, it should be able to produce a reasonable description of the photo.

Unsupervised Learning:

Unsupervised learning is the training of a machine using information that is neither classified nor labeled and allowing the algorithm to act on that information without guidance. Here the task of the machine is to group unsorted information according to similarities, patterns, and differences without any prior training of data.

Here we don't have any teacher to teach the machine, the machine is restricted to find the hidden structure in unlabeled data by our-self.

Unsupervised learning studies how systems can infer a function to describe a hidden structure from unlabeled data. The system doesn’t figure out the right output, but it explores the data and can draw inferences from datasets to describe hidden structures from unlabeled data.

An example of this includes clustering to create segments in a business’s user population. In this case, an unsupervised learning algorithm would probably create groups (or clusters) based on parameters that a human may not even consider.

Semi-Supervised Learning:

It is in between supervised and unsupervised learning since they use both labeled and unlabeled data for training — typically a small amount of labeled data and a large amount of unlabeled data. The systems that use this method are able to considerably improve learning accuracy.

source: researchgate

Reinforcement learning:

It is a learning method that interacts with its environment by producing actions and discovers errors or rewards. Trial and error search and delayed reward are the most relevant characteristics of reinforcement learning. This method allows machines and software agents to automatically determine the ideal behavior within a specific context in order to maximize its performance. Simple reward feedback is required for the agent to learn which action is best; this is known as the reinforcement signal.

Now we know what is AI and ML, let's go to our case study. How MNCs are getting benefited from using AI/ML.

In this case study, I am going to discuss how Google is getting benefitted using AI and ML.

Google uses AI and ML in every aspect of it, lets discuss one by one

Google Photos

  • The App will recognize the face on the photos based on the past user’s data, which is possible with the latest technology.
  • Once you snap a shot of someone, a known person whom you photographed earlier also, then the app will recognize his or her face and suggest sharing the photos with him or her.
  • In the case of grouped photos, the app will suggest the list of persons with whom you will share the photos based on the faces it recognized.
  • The Google App will also help you to delete the photos which are not good such as blurry photos. You can delete them easily on its suggestion.
  • Once you have provided the necessary permission to the app, the app will keep sending the photos automatically to the related person whose face it recognizes in the photos also share on connected social media platforms.

This has given a user-friendly experience to the users. The user can easily access his photos. He can also search for others by tapping on the faces. Google uses image processing in this app to determine the faces and sort them accordingly.

Google search:

Coming to google search it has been ruling the market of the search for a long time. In July 2020, online search engine Bing accounted for 6.43 percent of the global search market, while market leader Google had a market share of 86.86 percent. Chinese search engine Baidu’s market share was 0.68 percent.

Ever since the introduction of Google Search in 1997. Google has dominated the search engine market, maintaining an 86.86 percent market share as of July 2020. The majority of Google revenues are generated through advertising. The company has also expanded its services to mail, productivity tools, enterprise products, mobile devices, and other ventures. As a result, Google earned one of the highest tech company revenues in 2019 with roughly 160.74 billion U.S. dollars.

Google has been updating its algorithms in google search engines in such that the user gets his/her query resolved fast and efficiently. They are

Each algorithm used has AI and ML in them so as to make the search results effective.

Google handles over 40,000 queries per second. That is 3.5 billion searches per day.

Google Panda Algorithm:

Google Panda is one of the popular search algorithm updates that was released on 11th February 2011, with the aim of addressing the issue of spam. Before Panda was released, many companies were able to rank higher on the search results that produced low-quality content. These companies were called ‘Content Farms’. Their aim was to churn-out a large number of low-quality content every day and market them on Social Platforms and to earn a lot through ad campaigns.

Panda was an algorithm update that resolved this issue by giving a quality classification to every site. It was an AI-enabled algorithm that separated thin content from the quality content, which allowed sites with high quality and insightful content to rank higher providing better and more helpful search results.

Google Penguin Algorithm:

Google did not only struggled with thin content but also dealt with black hat link building techniques. A strategy, where a lot of low-quality links used to be added to the content to gain a higher rank in Google Search.

Penguin was rolled out with the aim to give higher rank to web pages that have good quality content and good quality incoming link. Penguin came out in several tweaks and refreshes, which helped almost 3% of the search results. Since 2016, Penguin is a part of Google’s core search algorithm and works in real-time.

Popular Machine learning technique, Pattern Detection plays a huge role in running the Penguin Algorithm as it allows the algorithm to separate content on the basis of low-quality attributes. This, in turn, allows those content to rank higher which actually deserves.

Hummingbird Algorithm:

While Panda and Penguin were minor updates in Google Search Algorithm, Hummingbird took the Google Search to a whole new level by introducing Artificial Intelligence technologies deep into it. Hummingbird came out with a capability of Semantic search and Natural Language Processing that allowed Google to understand the intent of the search query. For instance, if you type, “Michael Shoemaker”, Google fetches the following result:

‘Shoemaker’ is obviously the wrong spelling. However, Hummingbird realizes the intent and fetches the result that was expected. This clearly means that Semantic search enables the Google Search to parse the intent rather than the exact keyword. Semantic Search tries to match the SERPs with the language of searchers. The motive behind coming up with Hummingbird was to improve the voice search or conversational search.

RankBrain Algorithm

RankBrain Algorithm is another popular update that forms a part of Google’s Hummingbird algorithm. RankBrain is an Artificial Intelligence infused algorithm that fetches the best result for a query that is unknown to Google. Suppose you type “Orange” in Google Search. Now the keyword “Orange” may imply the fruit, or the French Telecommunication firm named “Orange”. In this case, you would get the result with mixed outputs.

RankBrain Algorithm is built on the Back Propagation Technique that is a standard method of training artificial neural networks. It fine-tunes the weights of a neural network on the basis of the rate of error obtained from the previous iteration. In the case of the RankBrain Algorithm, when an unknown query is searched, then the algorithm fetches the best fit results for the user. After which it compares the user’s rate of satisfaction and adds the phrase or keyword to the database if the rate of satisfaction is high.

Google store each search you have typed and store in their database so as to train the model in such a way that the results are accurate henceforth due to this training data. If we search unknown lines or content that google cant understand then it tries to show some relevant content related to that query.

Google Youtube:

Google uses deep learning today on its core services to provide more useful recommendations on Youtube.

Google Brain is behind the technology used here, which monitors and records our viewing habits as we stream content from their servers. Data already showed suggesting videos that viewers will want to watch next is key to keeping them hooked to the platform, and the ad bucks rolling in.

Deep neural networks were put to work studying and learning everything they could about viewers’ habits and preferences, and working out what would keep them glued to their screens.

The aim of deep learning at Google is to provide better video recommendations on YouTube, by studying viewers’ habits and preferences as they stream content, and working out what would keep them tuned in. Google already knew from the data that suggesting videos that viewers might want to watch next would keep them hooked, and keep those advertising dollars rolling in. Google Brain is once again the, well, brains behind this technology.

They use ML algorithms to find out the best ad for us and make us buy that product in some way later on. It also uses ML in child restrictions where the videos showed are restricted to children.

Google Translator:

Google Assistant speech recognition AI uses deep learning to understand spoken commands and questions, thanks to techniques developed by the Google Brain project. Google’s translation tool now also comes under the Google Brain umbrella and operates in a deep learning environment.

Google uses a deep learning model called sequence to sequence model.

It uses artificial neural networks that help google to accurately translate the given test based quote.

Google Neural Machine Translation (GNMT) is a neural machine translation (NMT) system developed by Google and introduced in November 2016, that uses an artificial neural network to increase fluency and accuracy in Google Translate.

Google translate supports 103 languages. Let's see some fun facts about it.

Google Translate translates more than 100 billion words per day.

More than 500 million people use Google Translate.

We can see Google has been actively participating in acquisitions related to AI and ML since 2009, It has acquired PittPatt in 2011, DeepMind in 2014, and its most recent recorded acquisition, Onward, in 2019.

With this graph, we can conclude the market for AI and ML.

“AI is one of the most profound things we’re working on as humanity. It’s more profound than fire or electricity,”

— — Alphabet Inc. CEO Sundar Pichai

Google also open-sourced its machine learning platform to the public on Nov 9, 2015. In November 2007, Google laid the groundwork to dominate the mobile market by releasing Android, an open ­source operating system for phones. Eight years later to the month, Android has an 80 percent market share, and Google is using the same trick — this time with artificial intelligence.

source: PWC

Thank you for reaching this point, Hope you have learned something from this article.

Do clap if you find this article useful and informative. Any suggestions are welcomed. Thank you





DevOps and Machine Learning Enthusiast