The "Asahi Shimbun DIALOG AI FORUM 2018" took place for five days from May 20 (Sun) to May 24 (Thurs), 2018, at BASE Q, a business collaboration hub in Tokyo Midtown Hibiya. This event considered possibilities for state-of-the-art technologies with a focus on artificial intelligence (AI), which is now in the phase of social implementation, and social issues. In their presentation titled "Future of AI Assisted Work," Mr. Koichi Hasegawa of Deloitte Tohmatsu Consulting LLC and Keizo Azuma of Fujitsu Limited took the podium and talked about changes currently occurring in their business fields and the latest case studies of technologies for solving social issues.
Trends in State-of-the-Art Technologies that Lead Corporate and Social Transformation
Mr. Hasegawa of Deloitte Tohmatsu Consulting LLC spoke under the theme of "Current Changes in Business Fields."
Why Do We Need Real-Time Data Processing Now?
Business have dramatically increased their data consumption in recent years. In the 1990s this mainly comprised descriptive analytics to explain what happened in the past. From around 2000, with the advancement of the so-called data-mining technology, diagnostic analytics have been implemented to understand causes using data diagnostically.
Recently, with the multiplication of technologies, analytics has expanded from analysis of the past to that of the future. In other words, predictive analytics that predicts "what will happen in the future" and further prescriptive analytics that prescribes "how to make a decision" using data have become mainstream.
The change occurring in current business fields is that predictive and prescriptive data analytics is expanding and that there is stronger demand than ever before for real-time processing to keep up with changing realities. The more we delve into "what we should do," the narrower the gap between the real world and the plan should be. This means the biggest challenge now is to frequently update the plan to perform analysis in a way that leads to real actions.
Real-time planning demands advanced computing capabilities. While computing performance has improved according to the so-called Moore's law (*), it remains unclear for how long this rate of growth will continue at the same pace. Consequently, expectations are escalating for a new architecture.
*: Moore's law, which refers to the number of transistors in semiconductor integrated circuits doubling in every two years, has acted as a guideline for the production and manufacturing of large-scale integrated circuits. This law was first proposed in a 1965 paper by Gordon Moore, one of the founders of Intel.
Trends in Quantum Computing Technology, Where Nations and Companies Battle for Supremacy
Recently, there has been a big breakthrough: quantum computing technology is gradually entering practical use. Quantum computing is a general term for computing technology that uses quantum phenomena and is even said to be destructive in that it completes calculations in only several seconds that would take conventional computers as long as three years. In 2015, Google announced that "as a result of operating a D-Wave quantum computer for two years, it proved to be 100 million times faster than a conventional computer." Such a new high-speed computing architecture has appeared on the scene and is garnering attention as a very effective technology for prescriptive analytics.
Quantum computing technology encompasses various methods, and major players in the computer industry from around the world are battling for supremacy by investing enormous amounts of money to develop quantum computing technology. Fujitsu has developed annealing technology that enables high-speed processing inspired by quantum computing technology, and launched a new architecture into practical use ahead of its peers.
Several nations have started long-term, large-scale projects worth tens to hundreds of billions of yen, and research and development are expected to accelerate further. The United States is investing more than 20 billion dollars a year in a project led by the Department of Defense. Meanwhile, European countries such as the United Kingdom and Denmark, and China, are also investing heavily, and other countries are paying attention to the technology as well. Although Japan has cutting-edge technologies, the investment in the 2018 budget is a paltry 2.2 billion yen. While national governments around the world have quickened their pace in developing quantum computing technology, most companies are yet to put such technology to practical use. It is predicted that approximately 20% of global top companies will examine the introduction of quantum computing technology into their business in 2023.
What Can Quantum Computers Do?
Note that quantum computing technology is not always perfect. It is important to identify areas that conventional computers cannot handle, and what is required now is to accurately explore what the new architecture can solve. To do this, some companies in various industries have already started PoC (Proof of Concept) activities for quantum computing technology.
Combinatorial optimization problems comprise one type that quantum computing can solve. Combinatorial optimization problems are latent in a wide variety of industries from finance to media, including optimizing logistics routes, simulating machine designs, analyzing proteins for pharmaceuticals, and optimizing advertisement portfolios in the media. Among various applications, it is vital to identify appropriate points for applying quantum computing.
Keys to Introducing State-of-the-art Technologies Are "Think Big," "Start Small," and "Scale Fast"
There are also problems in introducing state-of-the-art technologies. When compared with US companies and German companies, Japanese companies are not active in introducing AI and other leading-edge technologies. Three main factors slow down the introduction of state-of-the-art technologies: uncertainty of results, lack of ability to respond to new technologies, and resistance to change. Japanese companies in particular often contain organizational hurdles that hinder efforts to improve productivity using such technologies. If introducing avant-garde technologies is essential to secure competitiveness, how to implement trials is key.
We often suggest to "Think Big," "Start Small," and "Scale Fast." Draw a grand design, deepen the understanding of technology, and start small while understanding what you would like to do. In other words, it is important to have a medium-term vision, carry out a small and practical PoC, and implement a thorough approach for verification.
Digital Annealer - Fujitsu's New Architecture for Solving Business and Social Issues
Mr. Hasegawa and Fujitsu's Azuma talked in the panel discussion about how Fujitsu's technologies can be used to solve business and social problems.
Hasegawa: When working in business consulting, I often have opportunities to touch various technologies, but I've been thinking of quantum computing technology as a technology of the distant future. But the other day, Fujitsu announced the commercialization of its new technology "Digital Annealer," designed with inspiration from quantum phenomena. What are the features of this leading, cutting-edge technology?
Azuma: Digital Annealer is a technology that uses a digital circuit design inspired by quantum phenomena. The quantum world is a mysterious one where 0 and 1 exist simultaneously, which enables high-speed calculation. Fujitsu has continued research and development on quantum devices for more than 20 years. Developing Digital Annealer was the culmination of a unique question generated through interactions between a quantum researcher and computer researchers: "Can we do something related to quantum devices with digital computers?" So this development would not have been possible by the quantum device researcher alone nor only by computer researchers. The technology was invented only because a person straddled these two areas by chance.
Hasegawa: Fujitsu announced the development of Digital Annealer last year and has started full-scale operation this year at a cracking pace. What are your customers' reactions?
Azuma: We regularly release information and receive considerable feedback each time. For example, we received enquiries from customers in the financial industry through an investment portfolio case study, and those in the chemical industry through a case study on molecular similarity in drug discovery. Recently, we also received an enquiry from a customer in the logistics/distribution industry asking whether we could set up a project similar to our in-house factory flow-line optimization practice or a project developed based on the in-house practice.
Combinatorial optimization problems that we expect Digital Annealer to solve
Hasegawa: Optimization problems may sound unfamiliar to many of us, but they have existed for a long time. I think this kind of technology has given rise to a new challenge and provides a good opportunity to tackle them again. What kind of consultation requests are received from customers?
Azuma: In Japan, we often receive requests from customers that they want to speed up processing that takes a long time using software. On the other hand, there are many enquiries from overseas about the possibilities of doing something new with advanced innovative ideas, not things that have already been processed so far.
Hasegawa: Optimization is used not only in advanced fields, such as analysis of proteins in drug discovery, but also in fields familiar to us, such as analysis of professional baseball and professional soccer matches, behind the scenes. Actually, there are problems that require considerable time to process, even in fields closely linked with our lives. What do you think are promising markets for Digital Annealer?
Azuma: We focus on applying the optimization technique to optimize logistics flow lines, traffic volumes and routes, as well as shipping ports.
Fujitsu IT Products Limited Case Study: Digital Annealer improved productivity by reducing distance travelled within a warehouse by up to 45% (Japanese)
A case study of Fujitsu IT Products Limited applying Digital Annealer to optimize parts and rack layouts in a warehouse
Hasegawa: I think there are enormous opportunities in logistics and production sites. One of the advantages of Digital Annealer as a cloud service is that it is relatively easy to introduce by combining it with existing conventional mechanisms. What do you think Fujitsu will do to spread this technology in the future?
Azuma: First thing is to create a range of studies through co-creation with customers and in-house practices. The second is to make the technology easier to use by turning the case studies into APIs and standardizing them. The third is to expand the activities globally, rolling out Digital Annealer to North America.
Hasegawa: We have introduced Fujitsu's technology today in relation to the Deloitte High-Tech Competency Center (HTCC). The HTCC engages in activities focusing on applying technologies to solve problems in business and society by introducing cutting-edge solutions, combining business consulting with technology players like Fujitsu and academic research institutions. Currently, businesses are something that cannot be transformed by one company. Players in various areas make ecosystems and share information for new initiatives. Deloitte Tohmatsu Consulting has been working to connect leaders who are undertaking advanced efforts to create such communities. While promoting dialogues, we also hope to promote trials using state-of-the-art technologies.
Partner (Senior Manager at the time of this presentation)
Deloitte Tohmatsu Consulting LLC.
Head of AI Services Business Unit