Children are the hope for the development of humanity. At the same time, children are also disadvantaged and vulnerable groups and require special attention from the whole society, especially in the context that the current development of Artificial Intelligence (AI) is bringing profound and extensive impact to society in an unprecedented way.
All sectors of society should attach great importance to the impact of AI on children and develop responsible AI for the next generations. The development of AI should protect and promote the benefits of children, avoid depriving and harming children’s rights, and help realize the healthy growth of children.
| Children-centered values
The development of AI for children should align with the following values:
The development of AI should safeguard children’s dignity. The development of AI should value and respect children’s own thoughts, wishes, emotions, interests, self-esteem, etc., and should avoid harm to their personal dignity.
The development of AI should protect and promote the free development and diversified growth of children. AI should promote the development of children’s multiple intelligences and personalities, actively provide feedbacks to children’s curiosity, help stimulate children’s potential, and help guide children to form sound and scientific values.
The development of AI should treat and serve all children fairly and should not cause discrimination or harm to any child. The research and application of AI shall not make any distinction as to the child’s, his or her parent’s, legal guardians’, or other caregivers’ race, color, gender, language, religion, political or other opinions, nationality, or social origin, property, disability, birth or other status, as far as the fundamental rights of the child are concerned.
The development of AI should give priority to benefiting children. The research, design, development, deployment, and use of AI should prioritize the needs, rights, and interests of children, prioritize the adequate protection of children, and prioritize the promotion of children’s development.
| Protecting children’s rights
The development of AI should protect the following children’s rights:
The development of AI should help protect and promote children’s physical and mental safety. AI should help protect children from all forms of physical or mental violence, injury or abuse, neglect or negligent treatment, maltreatment or exploitation, including sexual abuse, and help combat child trafficking, indecency, and other crimes.
Physical and mental health
The development of AI should help protect and promote children’s physical and mental health. For instance, AI could help to enhance the ability to diagnose and treat childhood diseases, help to tackle children’s nutrition and health issues caused by poverty, hunger, environmental pollution, and other problems, and help to protect children’s normal cognition, character, ability, health habits, and behaviors.
The development of AI should protect children’s privacy in a much stricter manner. The collection of information on children should follow the principle of “legal, proper and necessary”, ensure their guardians’ informed consent, and avoid illegal collection and abuse of children’s information. AI systems should ensure that children, parents, legal guardians, or other caregivers have the rights to consent, refuse, erase data, revoke authorizations, etc.
The development of AI should help provide more inclusive, fairer, and quality education for children. The development of AI should help protect children’s right to education, help provide children with scientific, high-quality, and ethical educations, help children fully develop their personalities, talents, and abilities, and help avoid dangerous, coercive, unhealthy, and immoral educations.
Expression of will
The development of AI should help protect children’s right to freely express their opinions and wishes. Through appropriate channels, AI developers should seriously listen to, value, and consider the opinions of children. Children should be allowed to express their opinions and wishes freely when interacting with AI. Parents, legal guardians, or other caregivers can help children interact with AI while taking their wishes into account.
| Taking responsibilities
AI for children should undertake the following responsibilities:
The development of AI should uphold a responsible attitude towards the next generations, fully consider and try to reduce and avoid the potential ethical, legal, and social impacts and risks that AI might bring to children. The research, design, deployment, and use of AI should invite children and their parents, legal guardians, and other caregivers to participate in the discussion, actively respond to the attention and concerns from all sectors of society, and establish a timely and effective error correction mechanism.
The application of AI should adhere to strict and prudent principles, strive to control and minimize the potential risks to children. Considering that the influence of AI on children’s psychology, physiology, and behaviors is still to be studied, and children’s own thinking and behaviors are highly uncertain, AI technology and products for children should conform to higher standards and requirements in terms of maturity, robustness, reliability, controllability, safety and security, etc.
AI models, products, applications, and services should, on the basis of continuous efforts to improve their own transparency and interpretability, further consider the cognitive levels, self-needs, and expression abilities of children at different stages, provide the corresponding level of transparency and explanations, and provide children with effective feedback mechanisms and interaction methods.
Children and their parents, legal guardians, or other caregivers should be informed in a clear and understandable manner of the non-human interaction objects or non-real contents involving AI technology so as to ensure that they will not be misled or subject to the negative effects that may result from such AI applications.
Train and guide
Children, parents, legal guardians, or other caregivers should be actively guided to recognize and understand AI services properly. Multi-channel and multi-form training approaches can be applied to guide children, parents, legal guardians, or other caregivers to gain a reasonable knowledge of AI, and to avoid the possible blind faith, obsession, or misinterpretation of AI.
The continued benefits for children’s growth and development should be regarded as the development goal of AI for children. Attention should be paid to the uncertainty in the long-term application of AI on children, and short-sightedness that might bring in negative impacts on children in pursuit of short-term benefits should be avoided.
| Multi-stakeholder governance
The governance of AI for children should observe the following principles:
AI development policies and ethical norms to protect children’s rights and interests should be studied and formulated. Research on the potential impact of AI on children should be strengthened, forward-looking codes of conduct, laws and regulations, and technical specifications should be formulated, and long-term follow-up studies and periodic assessment mechanisms should be established. The healthy development of AI in the direction of protecting and promoting children’s rights and interests should be encouraged and supported.
Legal system improvement
Stakeholders of AI should consciously and strictly abide by the code of conduct, laws and regulations, and technical specifications related to children. AI legislations should pay attention to the impact of AI on children’s rights and interests, and should make it clearly and effectively reflected in the legal system. Both governance institutions and strict review and accountability mechanisms should be established to severely punish individuals and groups that abuse AI to infringe upon children’s rights and interests.
Children, parents, legal guardians, or other caregivers, as well as governments, relevant companies, civil society, and people from all walks of life, should be encouraged to participate in the discussion of the potential impact of AI on children. A multi-stakeholder collaboration and shared responsibility governance model should be encouraged, and the interdisciplinary, cross-domain, cross-sectoral, cross-organizational AI governance for children should be implemented.
Openness and sharing
A cross-regional, global, and comprehensive AI governance open cooperation platform should be established to share governance experience and methods of AI for children, to promote the common development of global governance of AI for children, and to empower the healthy development of children all over the world in the era of AI.
1. For the definition of “children”, it is recommended that reference be made to the United Nations Convention on the Rights of the Child (UNCRC) and the provisions of each country or region.
2. The description of values and children’s rights mentioned here is partially based on the text of the UNCRC.
Endorsement (Contact us if your organization want to support the endorsement and be involved in the implementation of the principles)
- Beijing Academy of Artificial Intelligence
- Institute of Automation, Chinese Academy of Sciences
- Institute of Artificial Intelligence, Tsinghua University
- Institute of AI International Governance, Tsinghua University
- Institute of Artificial Intelligence, Peking University
- Institute of Psychology, Chinese Academy of Sciences
- Institute of Computing Technology, Chinese Academy of Sciences
- AI Technology Innovation Strategic Alliance
- Tomorrow Advancing Life (TAL)
- Gaosi Education
- Tongxing Academy
- 360 Total Security