Light and Shadow of Artificial Intelligence

This article is written by Mr. Izumi Harada , Executive Fellow, Institute for International Socio-Economic Studies（NEC Group）

National Strategy needed based on global cooperation with EU
Artificial Intelligence is now said to be at the stage of the 3^rd boom.The backgrounds of this boom are attributable firstly to drastic progress and improvement of computing capability, secondly to a large amount of data created by sensors in IoT environment and created by smartphones widely spread, and thirdly to practical utilization of machine learning where AI itself gains knowledge from the huge amount of data available.

Specially after the emergence of “Deep Learning” as one of machine learning, new concepts and products such as autonomous driving, communication robot, IGO software actually defeating the world champion of IGO have been released one after the other. Deep learning contributes further to smoothen wide variety of products being shown and recommended by smartphones, to improve weather forecast accuracy enabling as a result less damages of agriculture by and more prepared protection against natural disasters and to improve the level of crime prediction to lower the crime rate.

Lately AI is utilized also in the recruitment of new staff at companies. In the near future such work as one cordially done by humans and specially supporting back-office work normally done by corporate staff will be replaced by AI.  White Paper on Economic and fiscal policy 2018 pointed out that the introduction of AI contributes to improve company’s productivity by 16%. AI has potential to alter substantially the corporate activities, industrial structure and even the style of social life.

Who to take responsibility?
There are, however, issues to be solved in the field of AI. Machine learning is based just on statistical analysis, and is not based on the actual reality. The results led by machine learning are only showing correlation. They are not showing causation. Correlation shows the interrelation of a plural number of variables, but, it does not necessarily mean that correlation can show causation as there might be possible cases for common factors or accidental coincidence.

Consequently, AI should not be considered accountable or responsible for its judgment or decision if only correlation exists. AI should, therefore, not be utilized for the area where safety is of prime importance or for the important decision affecting heavily people’s private life.

If input data for machine learning and algorism to define calculation process are wrong or biased, the results thereof could be wrong and biased.  If the results seem apparently wrong, it might not cause any serious problem, but, if not so apparently wrong, it is very difficult to verify whether the results are right or wrong and is virtually impossible to guarantee the correctness of the results.

As a result, AI might promote social discrimination and violate the privacy. In order to solve these issues Ministry of Internal Affairs and Communications are studying how the principle of transparency, the principle of ethics, the principle of safety and the principle of privacy should be maintained and observed in the development and utilization of AI.

Evolution in the military field
On the other side it is said that AI will evolve the military aspects. Specially when AI is mounted on robot or drone, new weapons are developed as AWS (Autonomous Weapon Systems) which will autonomously select the attacking target by machine learning and program, and further as LAWS (Lethal Autonomous Weapon Systems) which mainly target to kill humans. In this area substantial questions are raised and argued who will take legal responsibility of the killing of humans if caused by error of AI or caused by the hacking from outside.

Many European and American corporates and academies involved in AI declared already that they would not participate in the development, manufacturing, trading and use of this LAWS.
In August this year the international conference was held in Geneva to regulate the use of military weapons where Japanese Government made a declaration that Japan would not have any intension of developing LAWS, but, also expressed conservative view that hasty arrangement of regulation might damage the engineering progress in consumer products area. It is said that the military use of AI would change the global military balance and even the global political balance. In this area USA has dominant position currently but is chased up rapidly by Russia and China.

It is suggested and recommended that Japan should establish the cooperation with EU and establish its own national strategy and initiative in this field with thorough understanding that the development of AI related technology would change not only global economic power balance but also global political balance.

************************************
Izumi Harada
*President,Crisis & Risk Management Society of Japan
*Executive Fellow, Information Society Research Department, Institute for International Socio-Economic Studies
*Graduate School of Keio University
*Currently in service for Institute for International Socio-Economic Studies, moved from NEC Research Institute and the Association for the Promotion of International Trade, Japan
*62 years of age, born in Tokyo