Research and develop responsible artificial intelligence systems.

According to the Ministry of Science and Technology, following the general trend in the world, artificial intelligence systems are assessed to bring great benefits to people, society and the Vietnamese economy through supporting and solving difficult problems that people and communities are facing. Along with that process, it is necessary to research and have measures to minimize risks in the process of developing and using artificial intelligence and balancing related economic, ethical and legal factors. Therefore, specialized agencies need to research and develop standards and guidelines for orientation, even if they are soft and non-binding regulations.

In the guidelines, the Ministry of Science and Technology sets out nine principles for responsible research and development of artificial intelligence systems.

First, the spirit of cooperation and innovation: Developers need to consider the connectivity and interoperability between their AI systems and other AI systems by considering the diversity of the systems to enhance the benefits of the system, while enhancing coordination to control risks.

Second, transparency: AI systems that comply with this principle are often systems that can affect the life, body, privacy or property of users or related third parties. In this case, developers need to pay attention to the ability to clearly identify the inputs and outputs of the system as well as the related explainability based on the characteristics of the applied technology and how they are used to ensure the trust of society, including users.

Third, controllability: To assess the risks associated with the controllability of the system, developers need to conduct a pre-assessment of whether the system meets the relevant technical requirements and standards. One of the risk assessment methods is to test in a separate space with security and safety measures in place before putting it into practice.

Fourth, the safety principle: Developers need to ensure that the artificial intelligence system will not cause harm to the life, body or property of users or third parties, including through intermediaries.

Fifth, security principles: In addition to complying with documents, instructions and implementing information security measures as prescribed, developers need to pay attention to points such as: Reliability and resistance to physical attacks or accidents of the system; confidentiality; integrity; and availability of necessary information related to the information security of the system...

6. Privacy: Developers need to ensure that AI systems do not violate the privacy of users or third parties. Privacy includes spatial privacy (peace of mind in personal life), information privacy (personal data), and confidentiality of communications. Developers need to apply current regulations and guidelines; international standards and guidelines on privacy can be consulted.

Developers should conduct a pre-assessment of privacy risks and conduct a pre-assessment of privacy impacts. To the extent possible, take measures appropriate to the characteristics of the technology applied throughout the system development process to avoid privacy intrusions when put into use.

Seventh, respect for human rights and dignity: When developing artificial intelligence systems that involve humans, developers must take special care to respect the human rights and dignity of the individuals involved. To the extent possible, depending on the characteristics of the applied technology, developers should take measures to ensure that they do not cause discrimination or unfairness due to bias in the data when training the system.

Developers need to take precautions to ensure that the system does not violate human values ​​and social ethics according to the basic principles of Vietnam (for example, values ​​including patriotism, solidarity, self-reliance, loyalty, honesty, responsibility, discipline, creativity, etc.).

8th, user support: To support users, system developers need to pay attention to points such as: Creating ready-made interfaces to provide timely and appropriate information to help users make decisions and use conveniently; implementing measures to make the system easier to use for vulnerable people in society (the elderly, the disabled)...

9. Accountability: Developers need to be accountable for the systems they develop to ensure user trust. Specifically, they need to provide users with information to help them choose and use the system. In addition, to increase social acceptance of the systems, developers should do more: Provide users with information and descriptions of the technical characteristics of the systems they develop, algorithms, security mechanisms, etc.; listen to opinions and dialogue with stakeholders.

Source: https://nhandan.vn/nghien-cuu-phat-trien-cac-he-thong-tri-tue-nhan-tao-co-trach-nhiem-post814609.html