“ARCC”: An Ethical Framework for Artificial Intelligence

|TRI-EN 作者:jeffcao 2020-04-09


Pony Ma, Chairman and CEO of Tencent, proposed that the future development of AI needs to be available, reliable, comprehensible, and controllable


Further Explanation

Jason Si, Dean of Tencent Research Institute, translated and elaborated the four principles as “ARCC” (same pronunciation as ark)

In his opinion, just as the Noah’s Ark preserved the fire of human civilization, the healthy development of AI needs to be guaranteed by the “ethical ark”


Principle : AI should be available

Human development and well-being

Ensure AI is available to as many people as possible, to achieve inclusive and broadly-shared development, and avoid technology gap


Human-oriented approach

Respect human dignity, rights and freedoms, and cultural diversity


Human-computer symbiosis

Relation between AI and human is not an either-or relationship, on the contrary, AI can and should enhance human wisdom and creativity


Algorithmic fairness

Ethics by design (EBD): ensure that algorithm is reasonable, and data is accurate, up-to-date, complete, relevant, unbiased and representative, and take technical measures to identify, solve and eliminate bias

Formulate guidelines and principles on solving bias and discrimination, potential mechanisms include algorithmic transparency, quality review, impact assessment, algorithmic audit, supervision and review, ethical board, etc.


Principle: AI should be reliable

General requirements

AI should be safe and reliable, and capable of safeguarding against cyberattacks and other unintended consequences


Test and validation

Ensure AI systems go through vigorous test and validation, to achieve reasonable expectations of performance


Digital security, physical security, and political security


Privacy protection

Comply with privacy requirements, and safeguard against data abuse Privacy by design (PBD)


Principle: AI should be comprehensible

“Black-box” technology

Promote algorithmic transparency and algorithmic audit, to achieve understandable and explainable AI systems


Differential transparency

Different entity needs different transparency and information, and intellectual property, technical feature, and technical literacy should also be considered


Explanation rather than technological transparency

Provide explanation in respect of decisions assisted/made by AI systems where appropriate


Public engagement and exercise of individuals’ rights

Various ways of engagement: user feedback, user choice, user control, etc.; use the capabilities of AI systems to foster an equal empowerment and enhance public engagement

Respect individuals’ rights, such as data privacy, expression and information freedom, non-discrimination, etc.; challenge decisions assisted/made by AI systems; provide relief for victims in respect of AI-caused harms


Informational self-determination

Ensure individuals’ right to know, and provide users with sufficient information concerning AI system’s purpose, function, limitation, and impact


Principle: AI should be controllable Effective control by humans

Effective control by humans

Avoid endanger the interests of individuals or the overall interests of the human species


Risk Control

Ensure the benefits substantially outweigh the controllable risks, and take appropriate measures to safeguard against the risks


Precautionary principle

Ensure AGI/ASI that may appear in the future serves the interests of humanity


Application boundary

Define the boundary of AI application


To build trust in AI, we need a spectrum of rules, ethics is just the beginning

Light-touch rules e.g. social conventions, moral rules, self-regulation, etc.

Mandatory rules e.g. standards, regulations, etc.

Criminal law

International law


2022年,从引爆AI作画领域的DALL-E 2、Stable Diffusion等AI模型,到以ChatGPT为代表的接近人类水平的对话机器人,AIGC不断刷爆网络,其强大的内容生成能力给人们带来了巨大的震撼。