How is generative AI being used?

The expansion of generative AI services is unstoppable. According to a survey by ICT Research & Consulting, Inc., the number of people using generative AI services is expected to reach 19.24 million by the end of 2024.

The stock prices of GPU-related companies, led by NVIDIA, are growing unstoppable, and it is no exaggeration to say that the IT industry = generative AI-related business.

However, since GPUs, which are popular for generative AI, and CPUs used in computers that read and use them have completely different design concepts, it has been pointed out that the two may progress as completely different industries in the future. So what kind of generative AI services are Japanese people actually using?

The same survey by ICT Research & Consulting also includes the results of a web survey of 1,243 users who have used generative AI services within the past year. According to the survey, the most popular by far is OpenAI's ChatGPT. In fact, 18.3% of generative AI service users use it.

Next is Microsoft's Copilot with 8.9% and Google's Gemini with 5.4%. Excluding these top three services, the usage rate suddenly drops to less than 3%, and it seems that the top services are beginning to monopolize the services.

ChatGPT is likely due to the start-dash effect of being an early service and being well-known, but for the second and third place services, it seems that people started using them before they even realized it because they were automatically installed in existing office apps and Android smartphones. The fact that major platforms are strong even in up-and-coming generative AI services seems like a bleak dream.

On the other hand, there are actually some bright signs. The survey also looked at user satisfaction with each service, and the service that three out of four people said they were satisfied with was Adobe's Adobe Firefly. This service is a generative AI specialized in image and video processing that is popular among designers and other creators.

In the user survey mentioned above, only 1.2% of the total people used it. However, it has achieved this high level of satisfaction due to its excellent UI. It seems that if the quality of the service is improved, it can compete with super-large capital.

Microsoft's Copilot was developed using technology from OpenAI, which provided financial support. Generative AI technology can be incorporated into apps as various original services through API integration*1.

Generative AI-related services being rolled out in Japan are lagging behind in the development of their own large-scale LLMs, so it seems that this type of application development is more popular. In addition, mentor-like businesses that provide advice on how to use generative AI in business, such as Digirise Co., Ltd., featured on page 8 of this issue, are also prominent.

This may seem like the opposite of what was said on the previous page, but a survey released by the Ministry of Internal Affairs and Communications in July of this year showed that while the number of people using generative AI for work is steadily increasing, the proportion of people using generative AI personally is less than 10%.

In the United States, 46.3% and in China, 56.3% use generative AI services privately, and this figure is so low that it is hard to believe it is a developed country. The most common reason given was that over 40% of people said they "don't know how to use it." This suggests that AI mentors and AI consultants are in demand in Japanese companies.

Many people do not use AI because they believe it is not necessary for their lives, and Japan has the highest percentage of such people in the world. The Ministry of Internal Affairs and Communications has analyzed that Japanese companies are "carefully introducing AI, starting with internal tasks such as creating minutes," and are rapidly establishing safe and secure rules so as not to be left behind by innovation.

On the other hand, venture companies with few employees and insufficient human capital are taking this opportunity to actively introduce AI. Looking at the top five AI development companies in Japan, they are all start-ups that were founded after 2000, including "Appier Group," "PKSHA Technology," and "Brainpad."

In 2027, the size of the AI-related market in Japan alone is expected to exceed 1 trillion yen. In order to keep up with the times, it will be necessary to have a certain level of understanding of AI at the personal use stage.

*1: To expand functions by linking data with external applications and systems.

Communication

When generative AI first appeared, what surprised everyone was its natural conversational ability. Whereas previous chatbots could only give formulaic answers, a system that could respond by understanding the user's intentions as if it were a human showed a wide range of possibilities.

• DIALOGUE SUPPORTIn recent years, companies have been struggling with communication training, such as telephone response guidance for new employees. Young people, who no longer use telephone contact in general, tend to have a sense of discomfort when talking on the phone. Therefore, efforts have begun to have generative AI provide real-time conversation support.
• CHAT BOTChatbots are often seen on the inquiry windows of corporate websites. Until now, chatbots could only branch according to the flow of a given program, so there were times when they could not respond to the detailed requests of individual customers. However, by using generative AI, it is now possible to hear very detailed requests and connect them to the appropriate office.
• RECORDING AUDIO DATAMany people record the contents of conversations in frequent meetings, but listening to the recorded data and re-writing the minutes is a considerable effort. A generative AI has been released that can create real-time minutes and even summarize meetings just by importing audio data.

Analyze

A large amount of data is entered into Excel, and a graph that is easy to understand visually is created, and the analysis is carried out by interpreting the completed graph. However, when repeatedly analyzing data, it is difficult to gain new perspectives due to personal experience and preconceptions. Using generative AI may bring a new wind to analysis.

• Infrastructure development (INFRASTRUCTURE) It is difficult to check the deterioration of infrastructure that is not visible to the human eye, such as water pipes and gas pipes. By inputting the installation date and weather into AI and having it analyze the deterioration over time, infrastructure renewal work can be carried out efficiently with a limited budget. It is attracting attention as a means to solve the problems of Japan's declining population.
• Business improvement proposals (BUSINESS) Many managers are worried that even though they have allocated sufficient personnel and have them work on projects with enthusiasm, they are not seeing any sales. In such cases, bottlenecks often occur in places that cannot be noticed by the eyes and senses of the people involved, and more and more companies are relying on AI analysis to find the cause.
• MEDICAL A system has been developed in which medical AI, which has more knowledge than humans, can make detailed diagnoses simply by inputting diagnostic results and symptoms. Medical businesses are also busy with clerical work such as creating medical fee statements day and night. If AI could take over these clerical tasks, the burden on medical workers could be greatly reduced.
• DEMAND Various industries have experienced that there are items whose sales increase each season, but not many people can really understand in detail the reasons why those items sell. However, it is believed that AI can predict and derive new demand due to factors that humans overlook by analyzing the true causes.

Creative

The biggest feature of generative AI is that it is an AI that creates new things. Humans present ideas and the AI ​​does the actual work of creating them. This will become the norm in the future.

• Video generation (VIDEO) In recent years, the field of video generation has been particularly rapidly developing in the topic of generative AI. In response to a prompt about the kind of scene you want to see, it will quickly create a video that looks like live action or hand-drawn animation. It may still be difficult to sell it commercially as is, but it is now at a level where it can be used sufficiently for creating temporary images when making a movie. It is a field that is exciting to watch develop in the future.
• Text generation (TEXT) Overseas, it is no longer uncommon for books created by generative AI to appear in Amazon's novel sales rankings. Generative AI learns from story patterns from all over the world, so it is suitable for creating new stories that humans find interesting. In the business world, we will leave the creation of standard emails to AI.
• ILLUSTRATION When AI was first released to the public, it sometimes produced rather absurd illustrations due to discrepancies in recognition. However, it is now possible to create illustrations at a level that is acceptable for commercial use. The style can be freely drawn, from realistic to cartoonish. Each AI has its own specialty, and Microsoft now has an illustration generation function in Office.
• TRAVEL PLAN A service that has actually started to be commercialized is the creation of travel plans using AI. Just enter the spots you want to go to and the approximate time, and the AI ​​will make detailed suggestions on accommodation and create a time schedule to efficiently visit other nearby tourist spots. Due to the nature of AI, it has a weakness in that it is not good at suggesting minor spots with little information.

Related articles

Japan's Past and Present: Once at the Cutting Edge of AI Research and Development

The advancement of AI into society as seen from the successive Nobel Prizes awarded to AI researchers