AI Ethics: Enter the Dragon!

[Original document by Ministry of Science and Technology of the People’s Republic of China]

---

Overview:  On September 25, 2021, the National New Generation Artificial Intelligence Governance Professional Committee issued the “New Generation of Artificial Intelligence Code of Ethics” (hereinafter “the Code”). According to the Code, its aim is to “integrate ethics into the entire life cycle of artificial intelligence, and to engage in artificial intelligence related activities”.

---

Introduction

It’s quite mystifying to see a country as infamous as China globally for its AI ethics violations, come up with an Ethics Code for the world to sit up and take notice. Its violations list is endless, ranging from the use of Uighur-tracking facial recognition technology and the use of emotion detection software against them in its Xinjiang province, to its flouting of human rights norms and draconian manner of application of the social credit system. In fact, China’s reputation as a country with an appalling human rights track record has gone from bad to worse in the past couple of years or so. To come up with an ethics code in such a setting and at this point in time, is quite surprising, to say the least. Travel back in time to 2017, and you have the “New Generation Artificial Intelligence Development Plan” which outlines China’s policy to become the leading AI power by 2030. It is interesting to mention here that, according to the object part of the Code, a couple of its objectives includes, to thoroughly implement the “New Generation Artificial Intelligence Development Plan”, and detailed implementation of the “New Generation Artificial Intelligence Governance Principles”. Coming back to the Code, it contains 25 Articles divided into 6 chapters.

High-level overview

Below is a high-level overview of the Code:

Chapter One (“One” as given in the Code) (General Provisions) (Articles 1- 4) 

This chapter talks about integrating ethics and morals into the full life cycle of AI, promoting fairness and avoiding problems such as discrimination, privacy etc. What is interesting to note here is that the chapter not only talks about incorporating ethics, but also morals. Therefore, clarity on the definition of morals for the purpose of this Code and how the integration will take place becomes imperative. Further, the chapter states that, apart from applying to natural and legal persons, the Code, also applies to ‘other related institutions’ engaged in related activities such as artificial intelligence management, research and development, supply, and use. There is ambiguity surrounding the meaning of the term ‘other related institutions’, and without any elucidation, the same can have disastrous consequences to the entities concerned in today’s globalized world. Article 3 is one of the most important provisions of the Code, as it lays down some basic ethical norms to be followed by various AI related bodies, under six distinct heads. They are as follows;

• Enhance human well-being – This first heading lists out several high-sounding ethical guidelines to be followed. Some of them include, follow the common values of mankind, respect human rights and fundamental interests of mankind, improve people’s livelihood etc. One interesting norm is to promote harmony and friendship between man and machine. Whatever that means, it would be some task for the people associated with the field of AI to accomplish. Another ethical norm mentioned is “adhere to people-oriented”, which is extremely arduous to comprehend. 
• Promote fairness and justice – It talks about adherence to inclusiveness and inclusiveness, which again does not convey any meaning, whatsoever. The other ethical norms clubbed under this broad heading effectively protect the legitimate rights and interests of all relevant subjects, promote social fairness and justice and equal opportunities.
• Protect Privacy and safety – This head includes inter alia, fully respect the rights of personal information to know and consent, protect personal privacy and data security, information must not infringe on personal privacy etc. The above sounds more like clauses from a data protection statute. Although, incorporation of obligations concerning privacy and data protection seems like another layer of fortification for the rights-holders, but how far it will stay clear of not involving in an interpretation imbroglio with the recently passed Personal Information Protection Law (PIPL) will be one riveting duel to watch out for.
• Ensure controllability and credibility – It comprises ensuring that humans have full autonomous decision-making power, the right to choose whether to accept the services provided by artificial intelligence, the right to withdraw from the interaction with artificial intelligence at any time, and the right to suspend the operation of artificial intelligence systems at any time to ensure that artificial intelligence is always under human control. It is quite obvious that the above requirements (which entails some explanation), would prove extremely burdensome for the companies to follow. 
• Strengthen Responsibility – Insist that human beings are the ultimate responsible subject, clarify the responsibilities of stakeholders, introspect and self-discipline in all links of the artificial intelligence life cycle etc. are some of the ethical norms included under this head. Explications required include ‘ultimate responsible subject’, ‘introspect and self-discipline’ etc.
• Improve ethical literacy – Actively learn and popularize artificial intelligence ethics knowledge, deeply promote the practice of artificial intelligence ethical governance etc., are some of the ethical norms contained under this head.

Other chapters

Without going into the comprehensibility and the interpretability issues, the other chapters containing the various articles are as follows;

Chapter II (Management Standards) (Articles 5- 9)

The management standards are contained in this chapter. Some of them include “stay true to reality and rush for quick success in the process of strategic decision-making, correctly exercise power and use power”, etc.As it is obvious, the above requirements are extremely hard to comprehend and therefore, cannot be given effect to, in a meaningful manner.

Chapter III (R&D Specifications) (Articles 10- 13) 

Strengthen the awareness of self-discipline, improve data quality, strengthen security and transparency, and avoid prejudice and discrimination etc.

Chapter IV (Supply Specifications) (Articles 14- 17) 

Respect market rules, strengthen quality control, protect the rights and interests of users, and strengthen emergency protection etc.

Chapter 5 (“5” as given in the Code) (Specification) (Articles 18- 22)

Promote good faith use, avoid misuse and abuse, forbid illegal use of artificial intelligence products and services etc. 

Chapter VI (Organization and Implementation) (Articles 23- 25)

This chapter deals with the implementation aspect of the Code. It states that the specification is issued by the National New Generation Artificial Intelligence Governance Professional Committee, and is responsible for explaining and guiding its implementation. It further states that the management departments at all levels may formulate more specific ethical codes and related measures based on this code and combined with actual needs. Article 25 talks about coming into force of the specification on the date of promulgation, and its revision in due course according to the needs of economic and social development and the development of artificial intelligence.

Between the lines

Prima facie a proper drafting of the Code is conspicuous by its absence. In fact, it is very loosely drafted and seems not to have undergone any revision whatsoever, before publication. Further, it appears to have been passed in a hurry, the repercussions of which can be devastating. Apart from the syntactic and other grammatical gaffes, the Code brims with lofty ethical ideals, which are easy to prescribe but extremely difficult to implement in practice. Nevertheless, the burden now rests on the shoulders of the concerned authorities to provide more clarity, not only on the interpretation issues but also on the implementation aspects of the Code. Only time will tell as to whether the Chinese Government is able to deliver on the principles and standards enshrined in the Code. To sum up, China can draw inspiration from Robert Frost and his immortal lines, “And miles to go before I sleep”, as far as implementing the Code is concerned.