Frank Zhou Summer Sun
JunHe LLP, Beijing

Summer Sun
JunHe LLP, Beijing

Life science has been one of the most heated applications of artificial intelligence technologies (AI) in China. In 2022, while investments in the life science sector shrank significantly in China, companies that engage in ‘AI+’ sectors, such as ‘AI+drug discovery’, ‘AI+medical imaging’ and ‘AI+robotics’ continue to attract investments. China’s legal regime on AI is also evolving to address the potential ethical issues. This article introduces the principles China adopts to address ethical issues in ‘AI+life science’ applications and how China regulates certain specific ethical issues.

Principles of science and technology ethics

In October 2019, China established the National Science and Technology Ethics Commission (‘Commission’) to study and regulate ethical issues in ‘scientific research and technological development activities’, or STAs. ‘AI+life science’ is regarded as an integral part of STA. Therefore, ethical issues in ‘AI+life science’ are required to follow the same legal framework governing ethics in STA.

With the efforts of the Commission, the State Council and other authorities in China jointly published the Opinions on Strengthening the Governance of Science and Technology Ethics (‘Governance Opinion’) in March 2022, which lays out a comprehensive legal framework on science and technology ethics. The Governance Opinion provides five principles to regulate ethical issues in a STA, being: (1) it must help to advance human welfare; (2) it must respect the human right to life; (3) it must adhere to fairness and impartiality; (4) it must reasonably control risks; and (5) the process must be open and transparent.

Ethics review

In addition to laying out the principles, the Governance Opinion requires that STAs relating to human beings or experimental animals must be reviewed and approved in advance by the science and technology ethics committee of the sponsoring institution. An institution that engages in STAs such as life sciences, medicine and artificial intelligence shall establish a science and technology ethics committee if their research projects involve ‘sensitive areas of science and technology ethics’. If the sponsoring institution does not have such a committee, it needs to delegate the review to another institution which has such a committee. For international collaborations, it must comply with the ethics rules of each participating country.

The Drug Administrative Law in China has required that all clinical trials of drugs and medical devices be subject to an ethics review. Without clearance from the ethics review, clinical trials cannot proceed. However, traditional ethics reviews in clinical trials focus more on medical ethics rather than science and technology ethics, and the ethics committee are usually constituted by healthcare professionals. It remains unclear as to whether the science and technology ethics review can be consolidated into the current ethics review framework, and if it can, whether the ethics review committee shall be represented by more AI experts.

Algorithmic bias

Algorithmic bias is a unique ethical issue for AI applications. If a database fails to reasonably cover diverse groups such as women, minorities, the elderly or rural populations, the AI application trained by, or referencing to, this database may come to biased conclusions.

Chinese law has prohibited internet service providers from making recommendations based on algorithms that treat consumers differently according to their preferences and habits in transactions. The Personal Information Protection Law of China also requires that any data processor that makes automated decisions based on personal data must ensure the fairness and impartiality of the decisions. However, there is no uniform rule addressing the issue of algorithmic bias.

Human control

Telemedicine and AI-assisted diagnosis have been developing rapidly in China. In a traditional hospital, a patient chooses services based on the trust in the hospital and the doctor, but AI blurs these differences. Some clinics of traditional Chinese medicine have introduced the concept of a ‘virtual doctor’, which not only gives diagnosis through an algorithm, but also recommends prescriptions to the ‘real doctor’. Wide public concerns have led to the requirement in the Internet Diagnosis and Treatment Regulatory Rules (Trial Implementation) that in all internet diagnosis and treatment activities, AI must not make independent diagnosis or prescriptions, and the diagnosis and prescription must only be issued by human healthcare professionals.

Summary

Stephen Hawking once said, ‘our future is a race between the evolving technological forces and the wisdom to harness them’. AI and biotechnology are both double-edged swords. The two combined can be a powerful weapon for growth, or a source of risk. No doubt China has seen the risk and made efforts to harness it. However, its legal regime to regulate ethics issues in technology is still evolving and will need more time to mature.