Tell us what role your department plays in the AI business.

SuzakiThe AI Technologies R&D Department was newly established in OKI's April 2018 reorganization. We are continuing with the algorithm research of sensing technology, the technical development of data analytics and the development of AI dialogue and text analysis technology that were being carried out by our predecessor department. Additionally, we are responsible for overseeing and promoting AI related activities of the entire OKI group with such functions as: (1) collecting AI related information within the OKI group, (2) developing human resource to utilize AI, and (3) technical verification and support, which is a preliminary stage of PoC (Proof of Concept).

NishikawaThe IoT Solution-Business Promotion Department is in charge of developing common products handled by the entire ICT Business Group. As for implementation of AI technology, we work closely with the AI Technologies R&D Department, and for hardware development, we collaborate with the Fundamental Technology Center in our same business group.

What are the main achievements reached so far?

NishikawaSpecific product releases include the AI dialogue engine "Ladadie®" that utilizes natural language processing technology in July 2017 and the vibration waveform analysis software library "ForeWave" in October 2018, which is useful for facility maintenance of factories and railroads. For both products, we have achieved solid results co-creating with customers, and we are now incorporating them into solutions for manufacturing, distribution, retailing and financial institutions. As for "Ladadie," we believe that it can be applied to improving the efficiency of so-called general affairs operations in companies, which handle inquiries from many bases.

SuzakiIn the transportation sector, there is an actual case where we worked with a co-creation customer to apply AI technology to predict traffic congestion and time requirements using probe information. In that system, we incorporate real-time construction information and customer's experience values to come up with highly accurate predictions.
Recently, we are collaborating with external education/research institutions and companies to develop AI edge computing technology for implementing AI in the edge domain and core technology for enabling interworking between AIs.

What are the characteristics and distinguishing features of OKI's AI?

NishikawaOKI's strength lies in its ability to implement AI at the edge by making full use of the technologies cultivated in the field of communication signal processing. Rather than analyzing all the enormous amounts of data in the cloud, we think it is more efficient to process the data at the edge and provide feedback, and if necessary, interwork with the cloud. Currently we are pursuing "highly reliable analytics," "light-weight AI implementation" and "environmental resistance" to realize real-time high-speed AI, so that we can expand the application range to new fields such as fiber optic sensors and river monitoring.

Some customers say, "AI is difficult, and I don't understand it very well." How will you approach these customers?

NishikawaAlthough the AI is premised on data availability, it cannot be used unless it is clear "what it is you want to do and how you want to do about it." Even when using AI, listening to customer's specific requests and solving issues is a major premise. This is no different from conventional system development.

What is required of vendors who support customers' introduction/utilization of AI?

NishikawaThe most important point in evaluating AI is the accuracy of the output results. In order to improve accuracy, it is necessary to examine the handled data carefully and select the appropriate algorithm. Human skills are greatly involved in these tasks. I think increasing the number of personnel who are well versed in AI skills in addition to domain knowledge and the ability to understand customer issues is one of the challenges OKI must face.

SuzakiThe fact that human resource development is included in the role of the AI Technologies R&D Department is proof that OKI is strongly aware of its importance. As a specific initiative, we have developed an education and training program that is divided into five stages from the basics to application and practice. AI skill improvement has been promoted throughout the entire OKI group and we are starting to see results. In addition, we hold events such as "OKI AI Conference" that presents examples of in-house AI initiatives to raise employees' awareness of AI.

Tell us your goals and enthusiasm for the future.

NishikawaIn advancing AI implementation at the edge, we would like to realize the following three issues: "ready to use" (having a certain level of smartness even at the initial stage of introduction), "able to use it everywhere" (ensuring robustness(*)), and "become smarter automatically" (able to respond to future changes). The current AI system requires individual teaching in which data is input and taught for each case to create a model. However, if AI is to be incorporated into the edge, we would like to provide a system that is simple to introduce and immediately produces some degree of effect, thereby expanding the base of AI utilization. In general, AI is "stubborn and artisan." We will combine artisanship with customers' and partners' data and technologies to create an easy-to-use AI solution.

SuzakiWe will further refine the present AI technology, increase the number of personnel with good command of AI, continue to focus on providing better solutions for customers, and work on R&D of core technologies anticipated for the future. Actually, we are already studying the "next AI technology for deep learning." We will continue our research on not only packaging technologies but also core technologies.