×
Community Blog Building Smarter Apps - Learn How to Integrate AI

Building Smarter Apps - Learn How to Integrate AI

This article discusses AI and how to integrate AI with business applications for better outcomes. It also briefs about Alibaba Cloud's image search service.

By Alex Muchiri, an Alibaba Cloud MVP. He is the founder of Itesyl Technologies, a financial data and business banking solutions company.

Introduction

We are in the age of artificial intelligence and at the foundation is a new type of application, which has embedded AI capabilities. AI apps will help businesses to exploit new markets, offer differentiated services, and improve user experience. This capability will also help businesses better understand the behavior of customers and boost growth. We can anticipate new ways that customers can interact with enterprise systems and with other people as well. In an example scenario, it would be possible to offer personalized financial advice based on gathered insights and user data, or having chatbots that resolve customer complaints in minutes as well as generate detailed reports autonomously.

Much of the technology needed for such a future exists, and the remaining part is piecing them all together. Artificial intelligence will help automate many back-office tasks while reducing costs and improving productivity. It will also help enhance the security of IT systems and provide faster responses and resolutions. Evidently, this has a tremendous impact on business outcomes and customer satisfaction. As we advance into the future, the demand for AI capabilities in apps will be a necessity. Customer service will be at the center of these new changes.

Principles of Building AI-embedded Applications

For considering an application to be smart, it should be capable of making decisions independently. Usually, such a decision is contextual, meaning that it responds to different user inputs such as text or voice differently. At the core of this functionality is artificial neural networks, which synthesize information based on training and provide intelligent output. When a user provides an input, the neural network compares it to some generated graphs to determine the closest answer. Building a sophisticated AI model is a complex process that requires massive investment in research and development resources that may not be available to every app developer. However, there are artificial intelligence platforms that are either subscription-based or open-source that a developer could tap into using special APIs. Examples include Alibaba Cloud Intelligent Robot, Intelligence Brain, Image Search, Tensorflow, and IBM Watson.

Embedding AI into mobile applications can make apps popular and this helps with conversion. It is possible to go further and integrate virtual reality to result in some of the most advanced enterprise software. With that in mind, let's dive into some of the integrations of AI in applications.

Integrating AI Into Apps

We are yet to see the widespread implementation of AI in conventional applications. Of course, there are some popular AI apps such as Tmall Genie, Siri, Cortana, and Alexa, to mention a few. The inclusion of AI in apps depends on the desired goal, skill, and data. Predictably, most organizations do not have sufficient AI expertise. So, before discussing the specifics, let's observe some objectives that can be met using AI tools.

Objective

A question before embarking on any project would be, "what is the desired outcome?" Without a clear objective, it would be difficult to craft a useful AI solution that suits the needs of both the business and the end-user. However, there is a limit to what AI can do and that must be factored into the desired objective. So, what aspects could benefit from AI in an app?

  • Automate responses to customer
  • Fulfill orders
  • FAQ responses
  • Quick computations from backend systems
  • Search for items
  • Generating legal documents

Some people overestimate what AI can do. While such ideas may help propel AI to the next level of development, we still have a long way to get there. But, AI is becoming more sophisticated and may one day be advanced enough to select what API to call and even modify provided data to respond to user input. In these examples, we can already automate much of the back-office tasks.

Skillset

Many organizations understand that AI is important to their business but most have not acted upon yet. To begin with, data science talent is in short supply. AI problems are complex and costly to solve and this has led to an exciting new development: the rise of API-based platforms such as ET Brain, Intelligent Robot and Amazon's Alexa. But there is still a need to experiment with different models even with such platforms. Plug-and-play doesn't work with artificial intelligence as developers need to know what model to use for each particular use case. This is a profound challenge for developers because the whole process of making applications intelligent is a deviation from previous objectives of improving the user experience.

Data

Data is the final consideration when working with AI models: both, the training data and input data are equally important. Understanding the training data helps to improve the performance and efficiency of an AI tool. Actually, data is the most problematic component since each business solution requires a specific type of data but general-purpose AI solutions such as image search or speech recognition require generic tools that are easily available. The training datasets are also widely accessible from multiple sources freely. The quality and size of the available data thus play a major role in the models derived. Incomplete and flawed data will not yield desired results as the models will be defective.

Nonetheless, security is vital when incorporating AI into applications. Customer information needs to be protected, including personal information, private messages, login credentials, and confidential information. Some of these challenges could be overcome by using AI platforms that integrate intelligence into the application in real-time. That way, AI can yield better insights and enable deeper interactions, while securing data from leakage and misuse. Alibaba Cloud's intelligent platforms employ state of the art technology and remove the complexity out of AI, but as with any off-the-shelf solution, there is a need to look at customization to meet the project requirements.

What AI Can Do for Apps?

The top use-cases for AI in applications are conversational interfaces as well as speech and text analysis. Usually, machine learning algorithms running on a cloud platform are exposed via an API. The application can then consume to analyze speech, text, and image recognition.

Models help analyze patterns in text or speech and extract the intent or meaning. This could be sentiment or service that a user needs, such as frequently asked questions, generate a report or answer questions. A chatbot is a popular implementation of AI in applications to improve the conversational experience. Deep learning tools can help computers make decisions using training data.

Case Studies

In this section, we look at how artificial intelligence has been used in applications to create a better user experience, derive more value, and increase revenue generation.

Alibaba Cloud Image Search

Alibaba Cloud Image Search is a deep learning service for intelligent image recognition and search. The technology helps to search similar or identical images. This has implications for both businesses and the industry. It has two main features:

  • Product Image Search: With this feature, you obtain information about products that are similar or identical to the product in the search image.
  • Generic Image Search: With this feature you find images that have subjects or elements that are similar or identical to the search image.

Image search provides highly accurate results, quick responses, high throughput and performance, and flexibility to tailor the experience. According to Alibaba Cloud, in two typical scenarios, you could use the technology to solve a real problem.

Image Search for Online Shopping

Alibaba Group's C2C e-commerce platform, Taobao, has implemented an image search to improve product searches. Known as Pailitao, the feature takes a simple image of a product, without any textual product description it returns similar or identical products. This has improved the search experience for products through accurate results and less use of text descriptions. Launched in 2014, Pailitao has seen the number of daily unique visitors increasing steadily from several hundred to more than ten million per day. This is an indicator of high traction for the image search functionality in online shopping.

Image Search on Photo Websites

People are increasingly using their smartphones to access the internet. This led to the rise of social media and other photo-sharing platforms. These sites use sophisticated search algorithms, which are mostly based on text. It is usually problematic as a single search may yield thousands of results. The use of image search as the basis of a search engine can support the search of billions of images, as opposed to keywords. Overall, this is an efficient search method and improves the user experience.

Conclusion

The use of AI in business solutions will herald a new era in application development. It will make work easier and improve customer experience using intuitive tools and deep insights derived from data. Developers today must find new ways to embed AI in their applications, for user convenience and to target industry-specific problems. It is inevitable and will influence the development of technology in the coming years.

If you don't have an Alibaba Cloud account, sign up for an account today and try over 40 products for free worth up to $1300. Get Started with Alibaba Cloud to learn more.

References

0 0 0
Share on

Alex

53 posts | 5 followers

You may also like

Alex

53 posts | 5 followers

Related Products