Advisor, Fujitsu Laboratories Ltd.
The rapid evolution of smart devices, cloud services, networks and sensors is interconnecting businesses, people, information and infrastructure in new ways. Software is at the core of this emerging world of hyperconnectivity. How can we exploit software to deliver innovation? We sought the views of Yoshitaka Sakashita, advisor to Fujitsu Laboratories, on the role of software in realizing Fujitsu's vision of a Human Centric Intelligent Society.
A few years ago, the well-known software developer and investor Marc Andreessen made waves with an article entitled "Why Software Is Eating The World." Do you think his observation that companies and services need to restructure to survive in a world being devoured by software is becoming a reality?
Sakashita: Software is already used in virtually every field you can think of. It has taken a firm grip on every industry. One sign of this is that in the span of only a few years, the concept of "software-defined this, that and the other" is popping up all over the place. It started with software-defined networking, that is, using software to flexibly manage network configurations and functions, and been rapidly followed by software-defined datacenters, software-defined environments and even software-defined businesses.
On the other hand, another trend has become evident as companies and countries make their application programming interfaces (API) available in the public domain. Amazon in the U.S. has made its API available to the public, allowing countless businesses and individuals to access Amazon's functionality in their development and provision of new services.
An open interface that anyone can use is a big deal. Software's characteristics are making all kinds of data and services easier to use, connecting talented people, knowledge, information, and industries from all over the world. This is symptomatic of the tremendous changes software is bringing about in a hyperconnected world.
It seems that as software evolves, making inroads into new fields, we will have to use it in new ways in order unleash its tremendous potential. What should companies keep in mind if they want to create and deploy innovative products and services?
Sakashita: From a technical standpoint, there are many things they should do. For example, agile development is necessary to rapidly give form to ideas and system requirements. And to swiftly roll out services and functions in the market, test engineering is a must. DevOps*1 is also crucially important for developers to reflect knowledge gained by system operators. Furthermore, connection through an open interface means careful consideration of security issues.
What is most important is "loose coupling." When diverse systems, functions, and services are rigidly bound together, system maintenance becomes a problem. There is also a need to modify legacy systems and functions to facilitate the introduction of new technologies.
In the ideal ICT system architecture which Fujitsu proposes, we demonstrate how loose coupling can endow legacy systems and business processes with flexibility and agility so that they can adapt to new business styles and environments at the lowest possible cost. Specifically, using software as the platform for business process management, such as service coordination and data integration, allows us to easily link up multiple services or multiple business processes. Provided those services are loosely coupled, they won't adversely affect one another.
There used to be a lot of talk about Service Oriented Architecture (SOA) that combines service components to flexibly configure applications. In view of the emerging a hyperconnected world, loose coupling is a topic that resonates. But is loose coupling very similar to SOA?
Sakashita: I have been advocating the need for loose coupling since about 1990, when I was working on mainframe computers. In the intervening years, a host of concepts and software aimed at ensuring simplicity in connecting systems and flexibility in business processes, such as SOA, Business Process Management (BPM), and Business Rule Management (BRM), have made their mark in the field of information systems. These developments underline the importance of loose coupling.
In addition to the various rapidly evolving facets of ICT-smart devices and social media, sensors and cloud services-there are many legacy systems to deal with. So we should design loosely coupled architectures to allow flexible development and adoption of new ICT. This will encourage greater use of data to create value, which will support companies' competitiveness.
Use of big data is taking off worldwide, but it seems that further progress of software is required in order to unlock the tremendous potential of big data in a hyperconnected world.
Sakashita: Progress in big data analytics, as well as in machine learning and artificial intelligence, is absolutely vital. Fujitsu Laboratories approaches R&D from the perspectives of both social innovation and business innovation. The outcomes are beginning to be reflected in our products. One of those results is a Linked Open Data (LOD) application framework that treats the entire Internet as one gigantic database. It estimates the relatedness of vast amounts of data based on their meaning, and performs graph analysis, making it easier to narrow down useful data from a massive volume. In the realm of machine learning, we have collaborated with the National Institute of Informatics on developing artificial intelligence with the goal of enabling an artificial brain to score high marks on the test administered by the National Center for University Entrance Examinations by 2016 (the "Center Test"), and crossing the threshold required for admission to the University of Tokyo by 2021.
In the realm of AI, how does learning happen?
Sakashita: Machine learning and AI have been around for a long time, but there has been remarkable progress lately, in terms of both the volume of data that can be processed and the speed at which it is processed. It's a whole new world. Previously, in order to process data on computers, we created algorithms based on tree analyses and other methodologies to emulate human experts' thinking and decision-making processes. But nowadays computers create algorithms from massive amounts of data using analogical reasoning.
We have seen that software developed by an engineer who wasn't very familiar with the game of "shogi" can defeat a professional shogi player. Computers learn how to play shogi automatically, and it sometimes happens that the software can think of moves that wouldn't occur even to a shogi pro.
So we are not far from the point when computers will automatically learn a problem-solving approach without a human instructing them, and moreover, compared with a human they will be much more likely to come up with the optimum solution?
Sakashita: According to Gartner, "By 2017, 10% of computers will be learning rather than processing." Gartner also says "By 2024, at least 10% of activities potentially injurious to human life will require mandatory use of a nonoverridable "smart system." (source: Gartner "Predicts 2014: The Emerging Smart Machine Era" Tom Austin et al, 21 November 2013 Predicts 2014). Such forecasts may seem extreme, but I have no doubt that surprising things will happen in a hyperconnected world.
As software comes to the fore, will hardware play a diminishing role? Or is that an exaggeration?
Sakashita: The role of software is on an upward trajectory because it accelerates innovation. But I don't think that will necessarily result in a decline in the importance of hardware. I like to think of hardware and software as the two wheels of innovation, propelling us into a new world.
For example, the Internet of Things (IoT), where anything can communicate with virtually anything else, or a Cyber-Physical System (CPS), which unifies computer systems and the real world using ICT, creates new value precisely because it works with hardware-whether it be industrial machinery and automobiles, or smart devices and sensors. Examples that come to mind are self-driving cars, unmanned aerial vehicles (UAVs) and robots. Integrated with lightweight hardware, carrying cameras and sensors and capable of traveling long distances, the intelligence of software comes into its own-gathering and storing of all sorts of information, performing sophisticated analysis of that information, and making predictions based on the results. Fujitsu will continue to pursue R&D creating innovative products with hardware that embodies expertise and intelligence. As a result, step by step, we are transforming our vision of a Human Centric Intelligent Society into an everyday reality.
*1 DevOps: development concept unifying the efforts of development and operations personnel
Insight from our Experts
Diese Seite teilen