Hello everyone, My name is Kenshin Fujiwara and I am the CEO and founder of HACARUS Inc.
Through this series of blogs, I will discuss a wide range of topics of AI, from the history of AI to practical tips for the successful application of AI projects. I hope my blog posts will help you better understand AI and solve your business issues.
In the last blog, we discussed the potential of China in data collection. Today, we will continue to discuss differences in the use of AI between Japan and the rest of the world, in particular, in terms of R&D costs for AI development.
Personal Information Privacy vs Data Collection for AI Development
On the other side of the spectrum, in Japan, many people are reluctant to entrust their personal information to others. Previously, I introduced diagnostic support AI for the medical field. With this technology, it is possible to diagnose patients and find potential risk areas for cancer with a satisfactory degree of accuracy. However, concerns regarding personal information had to be addressed to proceed with the joint research. Even in the initial stages, it was necessary to go through a series of strict checks by the university’s ethics committee to gain permission to use the MRI image data.
While personal information privacy is important, it can be viewed as a disadvantage in the world of AI and deep learning research. For example, in the field of automated driving, Japan is lagging behind in terms of data volume.
Another example is from the Google Maps service. This application offers a street view that is built on data captured by cars equipped with 360-degree cameras and 3D laser devices. Google already has a large amount of road data in many parts of the world. This data is necessary for the development of self-driving technology.
Previously, I also introduced Waymo, a google-affiliated self-driving technology development company that has collected more than 20 million miles (32.2 million kilometers) of actual driving data since 2009. Efforts are being made to collect road data in Japan too. Recently, I have noticed more vehicles in the urban areas of Japan that are being used to collect data. However, U.S. companies like Google already have an overwhelming advantage in terms of data volume.
Compared to Japan, R&D Costs are More than Double Overseas
In AI development, Japan is not able to compete with foreign companies in terms of financial strength. According to data released in August 2021 by the National Institute of Science and Technology Policy, Japan spent only about 18 trillion yen (an increase of 0.2% from 2018) on R&D in science and technology.
This is a huge difference compared to the United States, which spent around 50 trillion yen (an increase of 8.9% from 2018), and China, which spent around 41 trillion yen (an increase of 4.4% from 2018).
There is also a big difference in the amount of money that is invested in startups like HACARUS in Japan. As a relatively successful startup, HACARUS has raised more than one billion yen. However, if you go to Silicon Valley, you will find numerous companies that have raised between 20 and 30 billion yen.
Availability of Human Resources
This lack of financial backing also affects the availability of human resources. This is because the technology industry is highly competitive and talented professionals often seek high levels of compensation.
When it comes to deep learning, some famous investors include Jeffery Hinton, Yoshua Bengio, and Yann LeCun. Currently, two of them are working at Facebook and Google.
In the west, as an AI specialist, even a researcher graduating with a Ph.D. or a young professional can earn more than 20 million yen a year. They are also typically offered stock options, however, this is not a common story for Japanese startups.
These differences in data resources and financial backing also carry over to technological capabilities. In general, ample funding and access to data will accelerate technological innovation.
These rapidly evolving environments will also attract talented people from all around the world, which will in turn attract additional investments. This is precisely the virtuous cycle that has been created in Silicon Valley, where technological capabilities are improved, driven by human and financial resources.
Sustainable AI, Sparse Modeling
While it is unfortunate, it seems that Japan is lagging too far behind to effectively enter the world of deep learning development. However, sparse modeling has the potential to compensate for the weaknesses of deep learning and propel Japan onto the global stage.
Recently, the world has begun to place a higher emphasis on carbon neutrality. Since sparse modeling has a very low energy consumption compared to deep learning, it has great advantages in this technological progress trend. When aiming to develop AI that is sustainable and contributes to society, the environment, and biodiversity, Japan can lead the world by promoting sparse modeling.
It is for these reasons that I honestly believe that AI technology that enables operation with small amounts of data, low power consumption, and high interpretability, such as sparse modeling, is the optimal path forward for Japan.
In the next blog, I will discuss how important accountability factor is in AI systems. For your updated knowledge and insight about AI technology, subscribe to our newsletter or visit HACARUS website https://hacarus.com.
