In the last couple of years, the Internet has become quite flooded with information and roundup articles about AI.
However, the current situation is that many people are self-solving with a vague understanding of AI.
For example, have you ever heard about the AI business that has been trending recently but didn’t get a good image of it, or have you ever wondered what kind of AI technology is actually used in AI business?
In this article, we will dispel the above-mentioned questions while mixing specific technologies that AI can currently implement and specific companies’ AI businesses.
In addition, we have organized the risks of introducing AI tools and specific examples of other companies’ AI businesses so that we can support the first step for managers who want to build an AI business at this stage. . By all means, please refer to everyone who is interested in AI.
Table of contents
- What is AI (artificial intelligence)?
- Why is business using AI currently attracting attention?
- What is deep learning that is often used in business these days?
- Market size of AI-based businesses
- Competitiveness in Japan’s AI business
- Relationship between AI mainstream technology and business
- Examples of overseas AI business
- Netflix
- Aeolus Robotics
- Japanese AI business case
- Mercari
- Sushiro
- Morinaga & Co., Ltd.
- How to start a business using AI
- Clarify the issues you want to solve with AI
- Decide how to deploy AI
- Collecting data to load into AI
- Enhance AI to a practical level
- implement and optimize
- Four risks to be aware of when doing AI business
- privacy risk
- safety risk
- Security risk
- Unexpected operation risk
- summary
What is AI (artificial intelligence)?
AI is an abbreviation for “Artificial Intelligence ” . AI can reproduce some of the actions performed by humans with software, so it can perform various tasks in business and daily life.
Why is business using AI currently attracting attention?
Why is AI-based business attracting so much attention today?
There are two main factors.
|
Due to these factors, AI is now everywhere in business.
What is deep learning that is often used in business these days?
Deep learning is “one of the machine learning technologies that enables flexible decision-making according to the situation by learning a large amount of data and automatically extracting common feature values”.
Deep learning is a part of machine learning, but it enables more accurate analysis than conventional machine learning, and can be said to be the spark of today’s AI boom.
In addition, AI, which has improved its accuracy through deep learning, is now being used across industries.
For example, image recognition, which is one of the specific AI technologies, is used for product inspection in the manufacturing industry, estimating the number of people at events, and cancer detection in medical care.
Market size of AI-based businesses
Currently, the market size of AI business in all industries in Japan is expected to reach 2,128.6 billion yen by 2030, which is approximately four times the 2018 level (530.1 billion yen) .
In addition, from 2021 onwards, the use of AI will expand as an important factor that plays a role in promoting DX (digital transformation) in Japanese companies.
Although the growth rate of AI-based businesses is not growing at a high rate every year, the business opportunities will continue to expand as technology advances, and it will become a market that will attract more and more attention.
Competitiveness in Japan’s AI business
As you can see from the graph above, the United States is currently the country that emits the most AI startups in the world. China was second and Israel was third, followed by developed countries. Japan was sixth, ahead of Germany and France.
From the above rankings, you can see that Japan is leading the world as an advanced country leading the AI business, although it is not as strong as the United States and China.
Relationship between AI mainstream technology and business
Next, let’s look at AI technology and business relationships. Have you ever wondered how AI is applied to business?
The table below shows specific examples of business challenges and how AI is actually used. We have organized it in an easy-to-understand manner for non-engineers who have never touched AI, so please take a look.
| Issue type | Commentary | example |
| classification | Predict which category a new input belongs to based on a predictive model. | Object recognition in images |
| continuous evaluation | Predict the value of time-series data at an unknown time point based on a prediction model. | Product sales forecast |
| clustering | Divide independent data into several categories based on their similarities and commonalities. | Customer classification based on customer information |
| Other optimizations | Predict the optimal output given a given input. | Calculation of optimal vehicle travel route for fuel and required time |
| Anomaly detection | Determine whether any input is abnormal. This task type can be viewed as a subclass of the classification (detecting inputs classified as “abnormal”). | Detecting machine abnormalities from machine vibration |
| Ranking | Sorting data according to certain criteria in some information retrieval system. | Ranking of recommended purchase items |
| Recommendation system | Present data that recommends the user to make a choice. Selection of recommended data is based on similar user behavior and user preference | Presentation of recommended purchase items |
| Generate data | Generate new data based on predictive models. | Generating songs belonging to a specific music genre |
Below is a graph summarizing the “problem type” and “which AI technology is being used” introduced in the table above.
If you want to do business using AI, it would be a good idea to refer to it.
| Issue type | Typical technology used |
| continuous evaluation | feedforward neural networks , linear regression |
| clustering | K-means, Affinity Propagation |
| Other optimizations | generative algorithm |
| Anomaly detection | 1 classification by support vector machine, k -nearest neighbors, neural network |
| Ranking | Ranking by support vector machine, neural network |
| Recommended system | collaborative filtering |
| Generate data | Generative Adversarial Networks, Hidden Markov Models |
Examples of overseas AI business
As you can see from the data on “competitiveness in the global AI business” introduced earlier, there is no doubt that the AI business in the United States is more active than in Japan.
Therefore, this time, we have picked up two AI business cases that are likely to be useful for Japanese companies and society, even in the United States, which is such an AI business powerhouse.
In particular, it will be helpful for managers of companies that sell products on their own e-commerce sites and apps, and those who are struggling with a shortage of human resources at nursing care facilities.
Netflix
Under this corona, the world’s leading streaming video distribution service provider, “Netflix”, which has doubled its net sales in the previous term and is still rising in stock prices, will be based on the AI recommendation system in the financial results for July to September 2020. , revealed that users’ content viewing is around 75%.
The AI recommendation system is a technology that utilizes user browsing history and purchase history data to search for users who have similar behavior, perform mutual analysis, and display products.
The system is expected to continue to increase business profits in the future, and is attracting the attention of many companies that sell products on the Internet.
Other well-known AI recommendation tools include Silver Egg Technology’s “Aigent Recommender” and Amazon’s “Amazon Personalize”.
Aeolus Robotics
Aeolus Robotics Corporation is a San Francisco-based company that manufactures and markets general-purpose AI-powered robots.
In recent years, the nursing care robots developed by the company have attracted a lot of attention for their ability to reduce the physical burden on staff in the nursing care business and to solve the labor shortage in advanced countries where aging is progressing.
Another major advantage is that it is easy to actually introduce and use robots in the field, and that they can be introduced at low cost.
Japanese AI business case
In recent years, an increasing number of companies in Japan have adopted AI to improve operational efficiency and increase profitability.
This time, we will introduce three examples of companies that are familiar to us.
Mercari
Mercari, a familiar flea market app, has introduced a photo search function that utilizes AI.
Traditionally, sellers had to research and fill out product information. However, with this function, it is now possible to automatically predict the product name, category, brand, etc. simply by taking a picture, without the seller having to enter product information one by one.
As a result, listing items on Mercari became easier, leading to an improved user experience.
” Machine learning ” and ” deep learning ” are applied to AI’s image search technology, and it will be one of the AI technologies that will be essential when selling various products on EC in the future.
Sushiro
Sushiro, a major conveyor belt sushi chain, has introduced a system that uses cameras equipped with edge AI to recognize the number of plates taken by the user and measure the number and price, thereby automating a series of accounting processes. I succeeded.
Edge AI means AI installed in an edge device, and edge means “edge”.
If Edge A is used, there is no need to transmit data to the I cloud environment, and real-time processing and processing in an offline environment are possible.
Morinaga & Co., Ltd.
Morinaga Seika, a major confectionery manufacturing company, has improved the operational efficiency of customer service by introducing a chatbot tool.
In addition, by being able to reach consumers who would like to ask a question, which they had no contact with until now, the number of inquiries has increased fourfold compared to before.
Famous chatbot tools include AI Messenger Chatbot and CAIWA Service VIII.
How to start a business using AI
Now that you have understood each AI technology and business case, next we will introduce how to launch an AI business.
Specifically, consider the following five steps.
- Clarify the issues you want to solve with AI
- Decide how to deploy AI
- Collecting data to load into AI
- Enhance AI to a practical level
- implement and optimize
Clarify the issues you want to solve with AI
When launching a business that utilizes AI, first clarify the issues you want to solve using AI as concretely as possible.
For example, it would be a good idea to divide the company into three divisions: the back office, sales and marketers, and the business strategy division, and clarify the AI technology that matches the issues faced by each division.
Decide how to deploy AI
After the issues are clarified, let’s decide how to introduce AI.
Basically, there are two choices: “Do you want to outsource the introduction of AI to other companies?” or “Do you want to develop your own AI?” If your company does not have human resources who can handle AI, do not hesitate to outsource AI to other companies.
Recently, the number of companies providing AI services is increasing, so there is a possibility that AI tools can be introduced at low cost.
Collecting data to load into AI
If you are going to incorporate the machine learning capabilities of AI into your business, collect relevant data.
The more data you collect, the better AI performance will be.
Enhance AI to a practical level
After collecting the data, improve the accuracy and performance of the model. For AI enhancements, use machine learning algorithms to train models, test them, and evaluate their performance.
If the performance evaluation of the test is low, let’s analyze whether the technology used in machine learning is correct or whether the loaded data is correct.
implement and optimize
After enhancing the AI model, let’s put it into practical use and verify its performance.
A common mistake is that some people think that “it’s over when AI can be put to practical use in business,” but that’s no good.
AI is a rapidly evolving technology. Therefore, it is important to always scan the latest data, review AI technology, and optimize AI.
Four risks to be aware of when doing AI business
From here, when fully introducing AI into business, it is necessary to prepare for the following four risks.
|
I will explain each in detail.
privacy risk
In order for AI to perform machine learning, it is necessary to collect and utilize a huge amount of data, but in this case, we must be especially careful of “leakage of personal information”.
Leakage of personal information not only damages the credibility of a company, but may also result in suspension of business or liability for compensation.
To prevent this from happening, it is important to constantly disclose information on personal information and privacy protection efforts to business partners and consumers, and to build a solid relationship of trust.
safety risk
Regarding safety risks, companies that develop their own AI are mainly targeted. Carefully inspect independently developed systems for potential physical or financial harm to users and third parties.
Security risk
In the future, as the AI industry develops, the risk of AI systems being interrupted by cyberattacks will become a serious issue.
For example, there is a risk that the production control system of the factory where the AI system is used may stop, or the AI-equipped car may become uncontrollable.
At the same time, attention may be focused on the cybersecurity field of AI in the future.
Unexpected operation risk
AI is a state-of-the-art technology that is still developing, and there is a risk of causing unexpected behavior.
Currently, there is an argument that it is necessary to create a system for monitoring and responding to AI by humans and AI when unexpected risks occur.
In fact, before introducing AI, it will be necessary to create a mechanism to monitor whether the AI system is operating normally.
summary
Thank you for reading the article to the end. In fact, this time I introduced how to build a business from the details of AI technology.
I would be very happy if you could clarify your knowledge about AI and use it as a reference for your future career and management.
