Cloud Artificial Intelligence (or How AWS Can Replace You)

Artificial Intelligence (AI) is an intelligence programmed into machines to allow them to behave in some activities similar to the human being. These activities can be for instance speech or image recognition, learning, planning or problem-solving. About month ago I presented you an article about machine learning (ML) tools available in AWS. Today I would like to focus on how to use these tools to enhance your business model. I will focus on use cases rather than on how these tools work.
The previous article was about machine learning, but today I introduced a new term which is artificial intelligence. Let me introduce another one – deep learning. All of these conditions are somehow related to each other and before we go further, I will present a short theoretical brief and will describe how they are linked together.

As mentioned before, artificial intelligence is a human intelligence put inside of the machines. Or at least this what we would like to be, as the real artificial intelligence does not exist. Yet. However, even if modern machines seem to be more and more intelligence, they do not think. They seem to be intelligent because they are loaded with a whole bunch of data, or rather big data, which they can analyze using sophisticated algorithms. This is where machine learning comes in. An ML model can analyze billions of examples to find a function that predicts output with a given input. The more advanced ML can be a deep learning (DL). One can say that deep learning is a machine learning enhanced with neural networks which are designed to classify information in the same way as a human brain does. What does it mean? It means that with large sets of training data, deep learning algorithms can be able to identify relationships between such elements as shapes, colors, words and more and create predictions or interpretations.

From this, an artificial intelligence is a bag of many tools, which combined together may act as a human. You should already know what AWS services are “intelligent”, but do you know, how can you use them in your product or business to replace a real employee? Or even not to replace. Replace is not a right word. I would say how to make them do things, that no employee over the world can do. Let’s take a closer look.

Machine Learning use cases

Machine learning can be used for analyzing a big data sets and making predictions based on these analyzes. What can it be used for in a real world? Imagine that your system can analyze customer feedback and product reviews and recommend actions for your product managers to enhance you offer or correct existing errors. If you know what to expect from your customers, you would be able to train your customer support to be ready for the customer demands.

Feedback and review analysis is not the only you can use machine learning for. Another typical use case is demand forecasting. You can leverage ML algorithms to make predictions basing on your products selling characteristics and let you know what products are seasonal or when there is an increased demand for the specific product. Having such information, you can plan production accordingly and be sure that you will never be lack of materials and will always be ready for the market demands.

If you already make such prediction, why not to go a step further and try to detect anomalies which do not conform to the expected patterns? You can analyze all your product reviews to find out that some of them which are inappropriate comes from the specific region or that among thousands of positive reviews there is some that point to the similar defects. You can also find customers who are not happy with your product or service and proactively engage them with promotions. You wouldn’t be normally able to detect such behaviors, but with artificial intelligence, it’s quite easy.

When talking about machine learning, there is one thing that has to be mentioned. It is content personalization. As one size does not fit all, you customers expect personalized content. With ML you can provide your customers a personalized experience. All you have to do is to build a unified profile that combines your data and third-party researchers and develop a personalization strategy. I’m pretty sure that you’ve already seen personalized offers which are based on your prior actions and predictive analytics models. Netflix and YouTube use it to offer you content which may be interested in you according to what you’ve seen before. Amazon and other shops can suggest you a specific product you may be interested in based not only on what you’ve already bought, but also on products or product categories you just browsed.

The AWS tools you can you for such predictive analysis are Amazon Machine Learning or, if you find it too difficult, Amazon SageMaker.

Time for Deep Learning

Besides of “simple” machine learning tools, Amazon provides few API-driven services which can bring intelligence to your application. You should have already heard about Rekognition, Lex, Comprehend, Polly etc. All of them involves deep learning layering algorithms to gain a greater understanding of the data. As mentioned before, these algorithms can identify relationships between elements. This leads to image and video classification, speech recognition, natural language understanding and recommendation engines. So, what can you do with such tools? Well, there are a lot of use cases. Let’s take look on some of them.

Image and video recognition are the most known of AI tools provided by AWS. With Amazon Rekognition you can detect objects, scenes, and activity in the video or image. What’s more interesting, you can identify a person in a photo or video using a repository of face images, you can analyze faces to find out for instance if the person is smiling. You can also track people through the video even if they are not visible. What possibilities does it give to you? Would you like to track your customers when visiting your store to find out what products they are most interested in or what paths do they choose when moving around? You can then decide where to put products on sale to be better visible and attractive to the customer. Why not compare a sell reports for specific products with the amount of customer passing by? Or maybe you want to know if any celebrities visit your shop? Why not. Just set up a camera and use Amazon Rekognition.

Another use case for Rekognition is unsafe content detection. If, for example, your web service provides images or videos uploaded by customers to other customers, you may want to identify a potentially unsafe or inappropriate content to control what you want to allow. You can also identify text in your images to search them using specific phrases.

A quite new service is an Amazon Comprehend, a natural language processing service to find insight and relationships in text. So what can you do with Comprehend? Let’s imagine that your products have thousands of reviews in the Internet. It’s impossible to analyze them manually but would you like to identify competitive products in them or would you like to know if the reviews are positive or negative? Or maybe you want to find out if your product is mentioned among your competitor’s names? Try to use Comprehend.

Another nice service using AI in the background is Amazon LEX, which allows you to build conversational interfaces into any application using voice and text. It’s the easiest way to build chatbots integrated with your Facebook account or call center (Amazon Connect) and use them for customer support. A chatbot is good enough for booking a room with a phone call and answering typical customer questions. What is beneficial, such chatbot never sleeps and will be cheaper than employing a real person.

It’s worth to mention, that Amazon LEX uses another AI technology inside. It is Amazon Polly, which translates text into voice and which of course can be used by you. What are the use cases? There are plenty of them. Any service that requires voice, can leverage Polly. Just to mention any dictionary applications, map navigations, learning platforms. One big advantage of using Polly is that you may make your application available for users with any reading disabilities.

Want more?

I hope I gave you few ideas of how you can use and benefit from using artificial intelligence in the cloud. Being honest, AI is the future. It will be implemented in more and more products and I bet that within next few years products without at least a little bit of AI inside will be on the margin. So if you want to be on top, you have to prepare now and implement AI wherever you can.

Unfortunately, it’s impossible to show all the features of Amazon’s AI services in so short article. Therefore, if you are interested in AI and want to use AWS cloud for that, you definitely should visit Amazon’s Machine Learning website which can be found here: https://aws.amazon.com/machine-learning/