[Trends in AI systems] Reasons why artificial intelligence has become popular and examples of familiar services

Table of contents of this article

  1. What is AI?
  2. AI system trends
  3. Examples of familiar services that utilize AI
  4. AI service development requires highly specialized skills
  5. Summary: Developing human resources who are strong in AI and transforming business

AI (artificial intelligence) has now come to be counted as a practical technology, and we are increasingly seeing it through familiar services. Many people may be thinking of incorporating AI into their business. To this end, it is important to understand what kind of technology AI is and to use it appropriately.

In this article, we will introduce the characteristics and trends of AI, as well as examples of its actual use. Please use it to develop and improve your own services.

What is AI?

In order to incorporate AI into business and achieve results, it is important to understand what kind of technology AI is and what it can be applied to. First, I will provide an overview of this type of AI.

Why AI has become popular

The term “AI” first became known in 1956, more than half a century ago. The term “AI (Artificial Intelligence)” was used at a study group called the “Dartmouth Conference” held in this year. Since then, various research and experiments have been conducted, but it took a long time for practical AI like today’s technology to appear.

It is said that the birth of more practical AI was triggered by an image recognition competition held in 2012. Thanks to dramatic improvements in computer performance and an innovative technology called “deep learning,” recognition accuracy has been achieved higher than ever before, and it has attracted attention. Even after the competition, deep learning-based AI technology continues to develop rapidly, and it has now become so widespread that it is widely applied to business.

AI classification

There is currently no clear definition of “what is AI?” However, there are two ways of thinking that can classify AI into two types, so I will introduce them.

The first is the idea of ​​classifying AI into the following two types from the perspective of “function”.

– General-purpose AI: Software that understands the situation and can perform any role
– Specialized AI: Software that specializes in a specific area and performs only a defined role

Second, AI is classified into the following two types from the perspective of “heart”.

– Strong AI: Software that has a mind and can act on its own judgment
– ​​Weak AI: Software that has no mind and only performs programmed actions

In fact, all modern AIs are classified as “specialized AI” or “weak AI.” “General-purpose AI” and “strong AI” have not yet been realized. It turns out that AI is not a one-size-fits-all technology, at least for now.

Mechanism and scope of application of AI

I explained that AI is not omnipotent, but there are no limits to its range of applications. This is because AI operations are based on the “learning” and “inference” functions described below.

– Learning: Accumulating knowledge in a specific area in advance
– Reasoning: Deriving rules, characteristics, etc. from an unknown situation based on knowledge

This means that although it is not possible to make one AI general-purpose, it is possible to prepare AI that has been trained individually to suit the purpose. Therefore, AI can be applied for various purposes. However, in order to build highly practical AI, a sufficient amount of high-quality data is required for learning.

AI system trends

AI system trends

In what fields is AI often used? Here, we will explain trends in the domestic AI market.

AI market size continues to expand

ITR Co., Ltd., which provides consulting and other services, has announced the results of a survey on eight major AI markets in Japan. The “8 major AI markets” are as follows.

– Image recognition
– Speech recognition
– Speech synthesis
– Text mining/knowledge utilization
– Translation
– Search/exploration
– Time series data analysis
– Machine learning platform

According to this survey, the market for “machine learning platforms” has seen particularly rapid growth. The scale has expanded by 44.0% in fiscal 2020 compared to the previous year. This result shows that the environment that will become the foundation for companies to incorporate AI into their business is rapidly becoming established.

The average annual growth rate for the domestic AI market as a whole from 2020 to 2025 is 18.7%. It is expected to reach 120 billion yen in fiscal 2025.

The use of image recognition is spreading to various fields

Image recognition is a technology that distinguishes what appears in an image. Although this was not easily possible with conventional technology, the practical application of AI has made mechanization and automation possible. Work that previously had to be done visually can now be done quickly and accurately by incorporating image recognition.

The use of image recognition is already widespread in various fields. For example, automation using image recognition is often used for inspection work and abnormality detection in factories. Other applications include suspicious person detection, facial recognition, and autonomous driving. AI-OCR, which can also read handwritten characters, is another example of image recognition applications.

The introduction of natural language processing is also progressing.

Natural language processing is a technology that analyzes natural sentences such as those written or spoken by people. It is easier to imagine this process if you think of it as a process of extracting meaning from text. As a result, it has become possible to have natural interactions similar to conversations between humans, even with AI partners.

As an example, so-called “chatbots” seem to be leading the way at the moment. It is often used to set up a customer support point on a website.

However, natural language processing is originally a technology that can be widely applied to services and businesses that handle text. In the future, we will likely see an increase in more advanced usage examples, such as document translation, search, and efficient use of knowledge to improve business efficiency.

Examples of familiar services that utilize AI

From here, we will introduce some familiar services in which AI is actually applied. These examples can also serve as references when developing and improving your own services using AI.

Determine quality by learning the know-how of connoisseurs

There are many jobs that require the know-how of veterans.

“TUNA SCOPE” is an attempt to replace such tasks with AI. This product uses image recognition AI to learn how to discern tuna, which is said to take 10 years to master. This is a solution that allows overseas factories to reproduce skilled work, which is becoming scarce in Japan.

Of course, there are some tasks that have value precisely because they are performed by people. However, in many cases, such image recognition solutions will be welcomed to improve inspection operations. This is because automating tasks that were previously performed visually, such as quality checks and preventing foreign material contamination, reduces human errors and reduces costs.

Identify products to streamline checkout operations

Image recognition is also being applied to improve cash register operations. This is because by identifying products through cameras, accounting tasks can be simplified.

Among such solutions, “BakeryScan” is targeted at bakery shops. Bread is a product with a different shape, and in some cases it is not possible to attach a price tag to it. Therefore, we made it possible to identify the type of bread using image recognition using AI. Stores that have introduced this product have seen benefits such as smoother checkout and fewer queues at the register.

Of course, the same thing is possible even if you are not a bakery shop. Depending on the size of the store, it may be possible to reduce the number of cashiers and staff, leading to cost savings. In addition to self-checkout, this is an area where development is progressing with a view to eventually realizing unmanned stores.

Search for information using natural phrases

Have you ever wondered why when searching for a website on Google, you can type sentences instead of words and still get relevant results? This is because the company’s search incorporates natural language processing.

By applying natural language processing, we can realize information retrieval in a more natural way. For example, this would be useful if you want to create an environment where you can quickly access information stored within your company. Specifically, one possible method would be to analyze various documents in advance and register them in a database so that even natural sentences can be searched.

Search solutions using natural language processing like this will continue to spread to many companies that want to make effective use of knowledge and streamline their operations.

Analyzing the emotions contained in sentences

The technology for determining people’s emotions from data is called sentiment analysis.

A typical method would be to analyze sentences using natural language processing and determine whether the meaning is “positive” or “negative.” This method is used in marketing, etc. They collect and analyze sentences written by users in the free response section of surveys and opinions on SNS, and use them as clues to understand how their products are being evaluated.

It is also possible to build an AI that can distinguish emotions not only from sentences but also from voices and facial expressions. This kind of AI is used for purposes such as reading the satisfaction level of customers who access call centers and checking the stress placed on operators.

AI service development requires highly specialized skills

AI service development requires highly specialized skills

AI is a highly specialized digital technology. Even companies that are accustomed to developing systems using IT may not be able to master it and incorporate it into their business right away. In order to achieve results in developing your own services that apply AI, you will need two main skills.

The first is engineering skills related to AI development. In addition to programming skills, the basics of mathematics and statistics, and specialized knowledge such as deep learning are required.

The second is the management skills needed to efficiently realize AI with the expected performance. It is important to understand the characteristics of AI and be able to lead the project using a method that fits the development.

Summary: Developing human resources who are strong in AI and transforming business

AI is one of the digital technologies that has developed dramatically in recent years. Knowing that it is not a one-size-fits-all technology, and using it appropriately based on its characteristics, it has the potential to be applied to a wide variety of services.

Leave a Comment