HOW AMAZON USES AI
Artificial Intelligence is a broad area of computer science that makes machines act like humans. The goal of AI is to mimic the human brain and function like humans. AI manifests itself smoothly and in many ways. We do not realize that we are using AI technology. If you ever asked Siri to play a song, you have interacted with an AI, if you ever asked for recommendations while browsing stuff on websites online, you have interacted with an AI. AI is designed so we do not realize that we are interacting with a machine.
Let’s talk about the biggest giant of the IT industry- Amazon. Amazon uses the technology of artificial intelligence very efficiently and smartly. It has reorganized itself with artificial intelligence in almost every part of the company. From the recommendation engine and Alexa to anticipatory shipping and dynamic shopping.
Amazon has used artificial intelligence with the approach of “flywheel”. A flywheel is an engineering tool that is used to conserve momentum and energy. For example, if we think of a potter’s wheel, it spins smoothly every time we add energy. Amazon’s flywheel approach is somewhat similar to this which states that the momentum of artificial intelligence has to be kept in constant motion. The entire organization is contributing their efforts and knowledge so that artificial intelligence can be spread from one part to another. The innovations made in one part of the organization are spread throughout so that all the spheres can learn and expand through them.
A very important part of AI is the “data”. Data is required to train the machine learning algorithms. Jeff Bezos made very clear in the year 2002 that data has to spread and be transferred to the whole organization rather than be held-in in one part.
Another approach of Amazon is “customer obsession”. Now let me explain how it works. You might not know this, but before ordering any product on the Amazon site, that product is already shipped to your city. Yes, you read it right! This is called “Anticipatory Shipping”. The question is, how does Amazon know what the customer is going to buy before the order is even placed? If I am an Amazon customer, then before placing my order how does Amazon know which shoes I am going to buy, which books excite me, which headphones I am going to add to my cart? Well, multiple artificial intelligence algorithms are running behind the scenes to identify which product I am most likely to buy based on whatever products I bought in the past or whatever products I am looking at on the Amazon site or whatever products I have kept in my cart or wishlist. Based on that, Amazon’s artificial intelligence algorithms predict what products I am going to purchase in the next couple of days. Now let’s assume that in a particular city, say Ranchi, there are 5 million Amazon users, so Amazon AI algorithms will identify what all products those 5 million Amazon users are likely to buy based on their previous purchase, search history, and cart items. AI algorithms predict that there is a possibility that in the next 10 to 15 days 800 people are going to buy ‘product A’. The AI algorithm knows that out of those 800 people only 400 people are likely to buy the product. Based on this past data it makes the decision to ship 400 packages of ‘product A’ to Ranchi city. So, that’s the reason why whenever we order anything on the Amazon site our order is immediately delivered to us.
Now, the next place where Amazon uses the concept of AI is in “Amazon Prime”. If you are a prime member of Amazon then you get different offers on all products and a promise from amazon to get quick delivery of a product within an estimated time. For example, you will get the delivery of the product within the next 11 hours or 24 hours. Now the question is, how does Amazon promise that they will be able to deliver a particular product so soon. The reason behind it is the estimations that are made by AI algorithms. AI algorithms predict behaviour based on anticipatory shipping, the number of products shipped, size of the product, availability of the delivery vehicle, and delivery boy. All these factors contribute to the estimation of when you will receive your product. Similarly, there are different AI algorithms working to provide personalized offers to you if you are a prime member.
Now let’s talk about dynamic pricing. Amazon AI algorithm decides the price of a product dynamically. When the product demand is high in the market, the Amazon AI algorithm increases the price of the product and when the demand is less, it decreases the price of the particular product to increase the sales. Amazon also uses AI to be competitive in the market. For example, if an “x” product is available in 900 rupees on amazon and in 850 rupees on other websites, say Flipkart, now what the AI algorithm does is, automatically decreases the product price to 800 or 815 to increase the product sale.
Artificial Intelligence algorithms are heavily used in product recommendation engines on Amazon. If you are viewing books on the Amazon site then just below the product description amazon recommends other books that you may like. Similarly, if you are viewing bags then it recommends you various other bags of almost the same price and size. By this Amazon is increasing its revenue by 35%.
Another very interesting and popular electronic device is “Amazon’s Alexa”. Have you ever used Amazon’s Alexa? I have! Once I asked Alexa that how has she been doing? And her reply gave me a hilarious shock. The device replied, “Main thik hoon, Mumbai ki bhaasha mein bole to ekdum jhakaass” [Translation: I am well, and as Mumbians say perfect!]. You may ask, how did it know the Mumbai slang word? And that it is popular in Mumbai? AI algorithms are behind this. Whatever we speak to Alexa, it gets converted into text and gets transferred to Amazon servers where various artificial intelligence algorithms are working and trying to understand the text and delivers the service accordingly to the Alexa device where it converts it into voice.
This way amazon is using AI very intelligently and efficiently. Starting from anticipatory shipping, amazon prime, dynamic pricing, product recommendation, and Alexa. Artificial Intelligence since its emergence has expanded exponentially. It has various domains such as Deep Learning, Machine Learning, Computer Vision, Image Processing, Natural Learning Processes, Expert Systems, Knowledge Base, and so on. Amazon’s CEO Jeff Bezos said in one of his interviews that “It is just the beginning of the AI applications and a lot more has to come”. It’s so exciting to see how new technologies make their place. We are on the edge of the golden era and a lot more fascinating things we will encounter in the future.
References:
1. https://www.youtube.com/watch?v=HeV_glcTl2M&list=LL&index=6
2. https://www.youtube.com/watch?v=Hfks4yaqlsg&list=LL&index=5
