AI Social Principles (AI社会原則) are the basic principles, set out in the Social Principles of Human-Centric AI adopted by the Japanese government in 2019, that society as a whole should share when putting AI to use.
Principle (2019 edition)
Aim
Human-centric principle
Use AI as a tool that supplements human abilities, without infringing basic human rights
Principle of education and literacy
Offer wide learning opportunities so that everyone can understand and use AI
Principle of privacy protection
Keep personal data from being used in ways the person does not want
Principle of ensuring security
Manage the risks that come with AI and build mechanisms for safe operation
Principle of fair competition
Prevent the resources AI needs from piling up in a few hands and distorting competition
Principle of fairness, accountability and transparency
Avoid unjust discrimination and be able to explain that AI is in use and why it reached a decision
Principle of innovation
Widen exchanges of research and people so that society can develop together with AI
The Social Principles of Human-Centric AI place three values at their core as the basic philosophy, then set out principles for realizing them. The 2019 edition has seven principles, framed as what society, especially legislative and administrative bodies such as the state, should keep in mind.
AnalogyIt is like a set of house rules the whole family agrees on. It is not for any one member, but shared so that everyone pulls in the same direction.
How to rememberIn the original Japanese title, the words for "human-centric" come first. The basic stance of putting people, not AI, at the center is written right into the name.
Exam points
ExamThe basic philosophy (基本理念) of the 2019 edition aims for a society that respects three values: human dignity, diversity and inclusion, and sustainability.
ExamThe human-centric principle (人間中心の原則) requires that the use of AI not infringe the basic human rights protected by the constitution and international norms.
ExamThe principle of innovation (イノベーションの原則) calls for collaboration among industry, academia and government and for people to move across organizations and borders, so as to build an environment in which people can develop together with AI.
AI Utilization Guidelines (AI利活用ガイドライン) are a handbook compiled by Japan's Ministry of Internal Affairs and Communications in 2019 that sets out what the side using AI (businesses offering services with AI, and the people who use them) should watch out for.
It takes the viewpoint of those who use AI rather than those who build it, sorting out the problems that can arise in use and how to prepare for them. Its backbone is the AI Utilization Principles, and for each principle it explains what users would do well to act on. It is offered as guidance that encourages businesses and users to act voluntarily.
AnalogyIt is like the safety notes in an appliance manual. They do not explain how the product is made, but how the person using it can avoid accidents.
How to rememberThe Japanese word 利活用 joins 利用 (use) and 活用 (put to good use). The name itself tells you the document is about using AI, not building it.
Exam points
ExamThe AI Utilization Guidelines (AI利活用ガイドライン) ask users to watch the quality of the data used for training and to take steps so that bias in the data does not carry over into decisions.
ExamThe AI Utilization Guidelines ask users themselves to foresee and prepare for harm or rights violations that may occur at the stage of using AI.
Mix-upThe AI Social Principles (AI社会原則) set out the ideals society as a whole aims for, while the AI Utilization Guidelines set out concrete points users should mind when using AI.
AI Utilization Principles (AI利活用原則) are the individual principles that make up the AI Utilization Guidelines and that those using AI are expected to follow.
AI Utilization Guidelines (2019 edition)
Principle of proper utilization
Principle of data quality
Principle of collaboration
Principle of safety
Principle of security
Principle of privacy
Principle of human dignity and individual autonomy
Principle of fairness
Principle of transparency
Principle of accountability
The 2019 edition of the guidelines lists ten principles, each with notes on what users should take care of. For example, the principle of human dignity and individual autonomy asks users to take care that AI does not unfairly manipulate people's decisions or emotions.
Exam points
ExamThe principle of safety (安全の原則) asks users to take care that the way AI operates does not harm the life, body or property of users or third parties.
Mix-upThe principle of transparency (透明性の原則) concerns being able to verify and explain AI's inputs, outputs and decisions, while the principle of accountability (アカウンタビリティの原則) concerns users' readiness to answer for their actions to the parties involved.
Mix-upThe AI Utilization Principles (AI利活用原則) are items listed inside the document called the AI Utilization Guidelines (AI利活用ガイドライン), not a separately published document.
Ethical Guidelines (倫理ガイドライン) is the umbrella term for guidance that sets out the values (such as human rights and fairness) to uphold when developing, providing or using AI.
Author
Example document
Government
Social Principles of Human-Centric AI
Academic society
JSAI Ethical Guidelines
Company
A company's own AI ethics policy
Governments, international organizations, academic societies, companies and others each write such guidance from their own standpoint. It asks, for example, that AI not create unjust discrimination, that privacy be protected, and that the reasons for a decision can be explained. It lacks the binding force of law and is a set of norms that those who make and use AI follow of their own accord.
Exam points
ExamOne of the values Ethical Guidelines (倫理ガイドライン) uphold is fairness (公平性), which asks that AI not produce results that unjustly treat people differently because of gender, origin or the like.
JSAI Ethical Guidelines (人工知能学会倫理指針) are the code of ethics adopted in 2017 by the Japanese Society for Artificial Intelligence (JSAI), an association of AI researchers and engineers, for its members to follow.
The 2017 guidelines consist of nine articles, asking members, for example, to contribute to the peace and welfare of humanity, comply with laws and regulations, and respect the privacy of others. The final article states that for AI itself to become a member of society or something close to one, it must be able to follow these guidelines just as members do. Whereas the government's principles address society as a whole, these are norms a professional body has set for itself.
AnalogyIt is like the rules of professional conduct a bar association sets for its members. Experts in a field draw for themselves the lines they must not cross.
How to rememberIts final article asks even AI itself to follow the guidelines, and that is the mark that sets it apart from the other guidance in this unit.
Exam points
Mix-upThe JSAI Ethical Guidelines (人工知能学会倫理指針) are one specific document covered by the umbrella term Ethical Guidelines (倫理ガイドライン), and their author is an academic society, not the government.
Ethical Guidelines vs. JSAI Ethical Guidelines
Aspect
Ethical Guidelines
JSAI Ethical Guidelines
Key differenceRefers to
Umbrella term for AI ethics guidance
One specific document
Author
Various: government, societies, companies
JSAI
Audience
Depends on the author
Society members (researchers and engineers)
Key points
The Social Principles of Human-Centric AI take three values as their basic philosophy, and the 2019 edition uses seven principles to describe how AI and society should relate.
The AI Utilization Principles set out, principle by principle, what AI users should take care of, such as safety and transparency.
Ethics guidance on AI is written by many kinds of bodies, including governments, academic societies and companies, and asks that values such as human rights and fairness be protected.
The JSAI Ethical Guidelines stand out for including an article that asks AI itself to follow them.