The conditions for the practical use of Artificial Intelligence (AI) technology are investigated. Now we are in the 3rd AI boom aiming at the active use of big data in the Internet. The AI technology, whose typical example is deep learning, is said to be g reat technological innovation and is expected to cause economic development. The famous Singularity Hypothesis attracts a lot of interests and insists that the AI smarter than human being will appear in time and will take human jobs. However, before talkin g about such a dream, we must explore the feasible conditions under which a desirable AI society can be realized, by noticing the features of current AI. Otherwise the boom will disappear in no distant future. The important feature of current AI is statist ical processing which yields probabilistic errors. Therefore the first condition consists in the establishment of maintenance technology to cope with these errors. In addition, we must recognize that, if we over trust the autonomy of AI, human being would be reduced to mechanical data and be led to an oppressive society without freedom. To prevent this, the second condition is necessary the innovation of information education which extends the concept of information from mechanical one to social/living on e. Unless these two conditions are met, the construction of human oriented information society will be unattainable.

View full abstract

The aim of this paper is to understand the social impact of Artificial Intelligence (AI) and robots both theoretically and empirically. Firstly, I introduce the theoretical framework I have developed for a deeper understanding of the social impact of AI and robots. Secondly, in order to address the differences between Western and Japanese perceptions and engagement with AI and robots, I demonstrate AI narratives within Japanese historical and social contexts since the 1920s. This work is done in collaboration with the Global AI narratives project at the Leverh ulme Centre for the Future of Intelligence, Universit y of Cambridge. Thirdly, I investigate Japanese engagement with AI and robots using the results of both qualitative and quantitative research. I introduce my two on going projects, firstly “Youth and project and secondly “Robots Engagement” project. Finally, I give some suggestion which I call “Human First Innovation” regarding the future of this field. I hope this study could help to create a better AI society together beyond techno orientalism wit hin the dichotomy between the West and the Rest.

View full abstract

The core technology of this third AI boom is the induction reasoning function based on a large amount of data. The result obtained by induction reasoning is not necessarily true. Also, since computers cannot handle meaning, the role of human beings is to impart meaning to inference results and to construct logic. In order to reflect the concept of human centric systems on inference results, it is necessary to deepen human centered ideals and clarify requirements and characteristics as a system. It is als o necessary to make rationality to being human centered systems.

View full abstract

This paper considers AI's Code of Ethics after accounting for the homogeneity/heterogen eity between human and machine. Based on the theory of neocybernetics, the difference between humans and machines is clearly shown and the ethical basis of “human oriented” society is confirmed. Next, the principle of fairness in the Draft AI Utilization Principles proposed in the conference “The Conference toward AI Network Society” is examined. The main point discussed in the principle of fairness in “Human Intervention is that judgments that affect human life require human beings to bear ethical responsibility for them, not pass it on to the machine. Regarding important decisions that affect human beings, which are an aggregate of the uniqueness of autopoietic systems, it is necessary for humans, who are also aggregates of autopoietic systems, to take ethical responsibility. Moreover, the principle of fairness should encompass inclusive thinking, which would prevent people from falling into social exclusion while increas ing social instability.

View full abstract

The essence of the cyber physical system is to create a space mapping real world into virtual world as digital twin. The cyber physical system senses various events occurring in the real world as data and accumulates the data in the virtual world. By analyzing, searching, and integrating this accumulated data using various statistics and artificial intelligence techniques, it is possible to discover various facts in real time. We will realize the cyber physical system as a new perception system. By realizing a new perceptual system called cyber physical system, we human beings will not only obtain a more objective overview of the real world, but also promote creative production activities. In other words, the cyber physical system including artificial intelligence technology enhances and supports human creativity while transforming from a society that satisfies a material desire to a society that satisfies a mental desire. A human centered system will be constructed with key production activities making full use of creativity.

View full abstract

What kind of near future does the artificial intelligence bring to us? Where will the Human oriented information system go to then? It is an ambitious essay to read this from a viewpoint of HOMO DEUS and the basic informatics.

View full abstract

The normality of classes is important for improving the capabilities of designing and maintaining application programs (APs). Two key requirements must be met to normalize classes: (1) the lifetime of each attribute value of an instance should be equal to that of the instance itself, and (2) each attribute value of an instance should have non transitive functional dependency on the primary key (PK) of the instance. However, in the ac tual design of APs, the attribute value of an instance is often copied to another instance by adding a new attribute that is transitively functionally dependent on the PK. We refer to this kind of normality destroying attribute as a “View Attribute.” Since the View Attribute destroys the normality of the class, it degrades the AP’s designability and maintainability. As a solution to this problem, this paper proposes a novel application architecture: (1) all classes of the AP’s entity layer objects are norma lized and designed by existence dependency classes, (2) the AP acquires View Attribute values by tracking the link with “multiplicity = 1,” and (3) a version management mechanism is implemented to manage the instances so that the attribute value can be obt ained at an arbitrary timing. Using the proposed architecture, the AP can set the pointer to the destination instances as the sourc e instance is created. As the result, the AP can easily and swiftly acquire the View Attribute values at any given timing. The experimental evaluation shows that an existence dependency class diagram can be made for an actual specification document of a public housing management system and that the View Attribute values appearing in the system’s approximately 40 application form s can be acquired by the proposed method.

View full abstract

A substantial increase in foreign visitors to Japan is expected over the next 10 years leading up to and following the 2020 Tokyo Olympic and Paralympic Games. Against this background, tourism information retrieval systems that are easy for foreign visitor s to use are desired. The usual approach to eliminating language disparities is to create multilingual interfaces for systems. However, it is difficult to cover the necessary number of languages through multilingualization. Accordingly, we develop an infor mation retrieval system that uses only methods that do not rely on language. By evaluating the operability an intelligibility of the system through experiments using participants, we demonstrate that it is possible to search tourist attractions and routes using methods that do not rely on language.

View full abstract

Traffic signal controllers plays a key role in control and prevention of traffic optical units. Most notably, installation of new traffic controller technology that calculates signal control parameters such as Cycle length, Split and Offset in real time based on traffic volume counted by traffic detector is actively promoted. However, this type controller wasn’t actually popularized by its h igh cost. On the other hand, a traditional traffic controller remains general used which controls optical units in accordance with the pattern table that combines signal control parameters for a long time. At this moment, this traditional type continues to use signal control parameters at time that traffic signals were installed. In this paper, an management support system which fits in the pattern table of signal control parameters with the traffic flow was proposed using probe information and proved this feasibility.

View full abstract

We propose the compound operation of a multiple-server backup system (MSBS) and a power-saving server system (PSS) that can operate independently configured as a power-saving, high-availability server system. The MSBS is based on multiple dynamic backup server systems (DBSSs), each of which manages only one target server. Usually, if two kinds of management programs are operated at the same time, both programs become complicated to prevent malfunctions. Here, we propose and adopt a system interface (SIF) to solve the complexity of the DBSS in compound operation, which is realized by allowing only the DBSS to request editing of the PSS configuration files. Rather than adding the file access and editing functions to the DBSS, we found that the structure was simpler when we added the SIF for this purpose.

View full abstract

In 2017 , readers used their iPhone to browse the email newsletters of compan y X and found that it displayed a product image of com pany Y. The iPhone's default mailer registers the received inline image in the HTML mail along with its cid in the cache. On the other hand, the cid of the inline images in the HTML mail delivered by mobyle t were not unique. Therefore, (1) When opening an HTML mail of company Y delivered by the system using mobylet on iPhone, the inline image is registered in the cache. (2) Another system utilizing mobylet by company X sends an image having the same cid as c ompany Y's inline image to the same iPhone in an HTML mail. (3) When opening the mail delivered by company X on the iPhone, Company X's HTML mail displays the inline image of Company Y registeredin the cache with the same cid, instead of the inline image of company X.

View full abstract