From October 27th (Wednesday) to 29th (Friday), 2021, the 2nd AI / Artificial Intelligence EXPO [Autumn] was held at Makuhari Messe.
At the AI/ Artificial Intelligence EXPO [Autumn], a wide range of AI-related services such as deep learning , machine learning , edge AI, natural language processing , image/ speech recognition , and dialogue AI were exhibited, with 16,428 visitors over the three days. Named.
In this article, we will report on the seminar “AI standards are finally being published ~ Points that AI people should know ~” by Mr. Ryoichi Sugimura of the National Institute of Advanced Industrial Science and Technology held at the venue on October 28th .
*Part of the lecture content has been omitted and the wording has been changed.
Table of Contents
- AI standards to be published | What does AI standardization mean in AI development?
- An unusual situation in which discussions on AI technology development and standardization are progressing in parallel
- AI that can be used in common anywhere in the world | Utilization of AI standardization in business
- Necessity of discussions based on ethical viewpoints
- AI/Artificial Intelligence EXPO [Autumn] |
- In conclusion
AI standards to be published | What does AI standardization mean in AI development?
An unusual situation in which discussions on AI technology development and standardization are progressing in parallel
Along with the progress in the use of AI in various industries, the movement toward AI standardization is gaining momentum around the world. Developing a framework to promote highly reliable AI is important both from a business perspective and from the perspective of improving the convenience of social life.
Discussions on standardization are being conducted in cooperation with organizations from around the world, and Japan is expected to play a major role.
Mr. Sugimura, who took the podium this time, is one of the people who are working to standardize AI in a committee called “SC42”, an international standardization organization, and are leading the world’s discussions.
At the beginning of his lecture, Mr. Sugimura addressed the audience by saying, “I would like business people to know the current state of AI standards.” Did.
Mr. Sugimura: Discussions on the international standardization of AI started three years ago, and we are about to reach the publication of the first document.
“Standards are decided” means that once international rules are decided, domestic standards cannot be decided arbitrarily. In other words, even if a rule is inconvenient for Japan, it cannot be changed later, so now that an international standard is being decided, Japan must be actively involved.
In general, the process leading up to standardization was often advanced (discussed) after the technology had sufficiently developed, but in the case of AI, standardization progressed in parallel with technological development, which is an unusual situation . I’m here. This is because they suddenly try to establish a standard even though it has not yet been considered as a technology. So those involved in the establishment of AI standardization are groping in the dark.
This standardization effort was approved in October 2017 by an international standardization body called JTC1 (Joint Technical Committee 1) and started in earnest in April 2018. International conferences are held twice a year. All meetings so far have been held in virtual space.
The number of standards currently being developed is extremely large, but about half of the standard areas in which Japan can directly participate in discussions. This is thanks to the companies, universities, and research institutes who are participating in SC42. At least from the start, I think Japan has been able to participate in the discussions in a good position.
Unless Japan does a good job of negotiating and arranging for the standardization of AI, even if it develops high-performance AI technology, it may not be recognized.
AI that can be used in common anywhere in the world | Utilization of AI standardization in business
Mr. Sugimura: Standardization has great significance from a business perspective. Recently, the word ” agnostic ” is often heard at conferences related to standardization . It means “agnostic” in English, but (in the world of software) it means “it works regardless of the specifications of a particular system”. For example, “platform agnostic”. Apart from that, you can run the same specifications on different machines, whether it’s Apple’s OS or Microsoft’s OS. It doesn’t matter what browser you use to search.
Creating something that can be used in any medium and in any world is called the agnostic approach. We are discussing the standardization of AI so that it can be used in various industries, whether it is medical, web marketing, or architecture.
Standardization areas in which Japan is involved include, for example, standardization of AI terminology and AI governance, AI quality, and standardization of functional safety. It is an idea that we should have a common understanding of what kind of life cycle AI has because it is AI with variations in functions and performance. These things are being created mainly in Japan so that they can be disseminated as international standards.
Once the AI standard is established, we will tell our customers that “AI is seen in this way even in international standards” and “We need to think about AI not only after purchase but also after delivery.” I think it will be the material to explain.
Necessity of discussions based on ethical viewpoints
Mr. Sugimura: I will repeat myself, but from now on we will move to capture a large market. We must create standards from the perspective of industrial development and consumer protection.
At the end of my talk, there is an ethical issue that must be considered. AI has the advantage of freeing people from boring and dangerous work, but using AI increases the number of things that we have to think about and may lead to the problem of discrimination, so we need to think carefully. should not do. We are proceeding with the project while talking with various researchers such as Professor Ema of the University of Tokyo (who is discussing AI governance), so we ask for your continued guidance and encouragement.
At AI EXPO [Autumn], in addition to Mr. Sugimura’s seminar introduced above, Mr. Yutaka Matsuo, a leading researcher in the field of AI, Mr. Noguchi of the Japan Deep Learning Association, and RIKEN’s Advanced Intelligence Comprehensive Research Mr. Sugiyama, the director of the center, and other gorgeous speakers presented the latest research and development status and business use cases.
AI/Artificial Intelligence EXPO [Autumn] |
At the AI/ Artificial Intelligence EXPO [Autumn], many services promoting DX were exhibited, and every booth was crowded with many business negotiations. We could see the high level of attention paid to AI, which is the core technology for promoting DX.
Tools for which demand is increasing due to the influence of the new coronavirus have also attracted attention. For example, many visitors gathered at booths for automatic transcription and automatic translation tools for online conferences.
Once the AI standards are published, for example, the quality characteristics of AI will be defined, and the overall picture of AI governance will be shown by building a basis for discussion. However, even if guidelines and standards are issued, it is meaningless if we do not know how to use them in the field. I hope that the “AI standard” that is utilized in business and research and development will permeate.
In addition, the acceleration of DX triggered by the new coronavirus is expected to continue the digital shift of companies. Taking the AI / Artificial Intelligence Expo as an opportunity, various companies will develop AI business utilization. AINOW will continue to post examples of AI utilization and the latest information on DX, so please check it out.