With the wave of digitalization sweeping the world today, algorithm technology has become an important engine for social progress and economic development. Whether in the fields of finance, healthcare, transportation or social media, the wide application of algorithms is profoundly changing people’s lifestyles and work patterns. At the same time, it has also brought a series of governance challenges, courier boys, takeaway riders, online car drivers, etc. have become “trapped in the algorithm” in the industry groups, attracting public attention. How to find a balance between promoting technological innovation and effective regulation has become an important issue for governments and industries. Recently, Deng Honglin, Associate Professor of our school, published an article in Jiefang Daily and Shangguan News, in which he put forward his insights on the main challenges, promotion paths, and policy recommendations of algorithmic governance. The following is the full text of the opinion.

With the wave of digitalization sweeping the world today, algorithm technology has become an important engine for social progress and economic development. Whether in the fields of finance, healthcare, transportation or social media, the wide application of algorithms is profoundly changing people’s lifestyles and work patterns. However, with the rapid development of algorithmic technology, the governance challenges it brings are also becoming more and more prominent, such as courier boys, takeaway riders, online car drivers and other industry groups “trapped in the algorithm”, which has aroused public concern.

Recently, four departments, including the Secretariat Bureau of the Office of the Central Committee for Network Security and Informatization, have launched the “Clear – Typical Problems of Algorithmic Governance on Network Platforms” special action, aiming to promote the normalization of algorithmic governance through strengthening supervision, improving laws and regulations, and other measures. This initiative is not only a response to the current situation of algorithm governance, but also a positive exploration of the healthy development of algorithms in the future. How to find a balance between promoting technological innovation and effective regulation has become an important issue for governments and the industry.

Key Challenges of Algorithmic Governance

1. Lagging technological development and regulatory progress

The rapid pace of technological development and regulatory lag is one of the main challenges facing algorithmic governance. Artificial intelligence algorithm technology itself has a high degree of complexity and rapid development characteristics, most of the algorithms used by Internet platforms are often closed-source, and their decision-making process is “black-boxed”, which makes it difficult for outsiders to understand their internal operation mechanism. This opacity also makes it difficult for regulators to identify and respond to potential risks posed by algorithms in a timely manner. For example, in the healthcare sector, algorithms are increasingly used in disease diagnosis and treatment recommendation. However, it is often difficult for doctors and patients to understand the decision-making process of algorithms, and if the algorithms are widely used without adequate regulation and validation, they may lead to misdiagnosis or omission of diagnosis, which will affect the effectiveness of treatment and safety of patients.

At the same time, the algorithm’s rapid iteration and ability to learn on its own makes its rate of innovation far exceed the rate of updating of existing regulatory tools and laws and regulations. The existing legal framework has failed to keep up with the pace of technological development, and many emerging technological forms and risks have not yet been included in the scope of regulation, resulting in regulatory measures that fail to effectively cover new issues in the application of algorithms. Therefore, how to establish a regulatory system that keeps pace with the times has become an urgent task.

2. Imbalance between platform responsibility and commercial interests

When designing algorithms, many platforms tend to focus mainly on improving traffic and user stickiness, and this commercial interest-oriented optimization sometimes leads to negligence of social responsibility and ethical norms. Commercial drive may lead to algorithms generating some social risks in the optimization process, such as discrimination, bias or misleading consumers, which have not been given enough attention. For example, recommendation algorithms on social media platforms may exacerbate the polarization of information, creating an “echo chamber effect” whereby users are exposed mainly to information that corresponds to their own views, thus affecting the diversity of public opinion. This phenomenon not only weakens the public’s exposure to multiple viewpoints, but also may further deepen misunderstanding and confrontation between different groups.

3. Public trust and lack of social consensus

The public has extremely high expectations for the transparency, rationality and fairness of algorithmic decision-making. However, the complexity and opacity of algorithms make it difficult for the public to trust their decision-making process. Especially when it comes to personal privacy and data security, the public’s concerns and questions about algorithms are even more pronounced. If the public lacks trust in algorithms and their governance mechanisms, it may lead to difficulties in the implementation of relevant policies and measures, or even resistance, which in turn affects the effectiveness of governance. Therefore, enhancing public recognition and trust in algorithm governance has become an important prerequisite for promoting the normalization of algorithm governance.

Algorithmic governance to promote the path

1、Sound legal system and regulatory norms

In response to the above challenges, it is first necessary to strengthen the construction of laws and regulations. Specialized laws and regulations should be formulated for the field of algorithms to clarify the requirements for algorithm transparency, user data protection and platform responsibility, providing a solid legal basis for governance. These laws and regulations should cover all aspects of algorithm design, development, application and evaluation to ensure that algorithms operate in a legal and compliant environment. At the same time, the laws and regulations should be flexible and able to be adjusted in a timely manner according to changes in technological development and social needs to ensure the effectiveness of regulation. By establishing a sound legal framework, the application of algorithms can be effectively constrained to ensure that they operate in a legally compliant environment.

2. Promote platform self-regulation and industry standardization

The government can encourage enterprises in the industry to work together to formulate technical standards and codes of ethics, and promote the formation of a self-regulatory mechanism to ensure that platforms follow ethical and social responsibilities while operating commercially. In addition, through the establishment of industry associations and technology alliances, cooperation and communication within the industry can be promoted to jointly address the challenges in algorithmic governance. Such industry self-regulation can not only make up for the possible lag and limitations of government regulation, but also enhance the social image and credibility of the industry as a whole. Enterprises in the industry should realize that sustainable development can only be achieved on the premise of following ethics and social responsibility.

3. Enhance public education and participation of all people

The government, in collaboration with educational institutions, can improve the public’s understanding of algorithms through publicity and education to make them realize how algorithms affect decision-making, especially in terms of data privacy and security. At the same time, the government and platforms can establish a communication mechanism with the public, listen to public opinions, and enhance public recognition and trust in algorithm governance. Increased public participation can effectively promote the transparency and fairness of algorithmic governance, thereby enhancing social acceptance of algorithmic technologies. In addition, interactive activities such as public consultations and hearings can provide the public with a more in-depth understanding of the application scenarios and potential risks of algorithms, which can lead to the formation of a consensus and promote the effective implementation of algorithmic governance.

Policy Recommendations for Algorithmic Governance

1. Improve algorithm transparency and interpretability

To promote the normalization of algorithm governance, it is first necessary to improve the transparency and interpretability of algorithms. Relevant regulations should require algorithm developers to make clear the principles of algorithm design, data sources and training process, so as to enhance the transparency of algorithms. In addition, the interpretability of algorithms can be improved through technical means, so that users can understand the basis of algorithmic recommendations and decision-making logic, and reduce the risk of privacy leakage and misinformation brought about by algorithms. Transparent algorithms not only help enhance public trust, but also provide necessary information support for regulation. Especially in areas of public interest, such as healthcare, finance and public safety, the transparency and interpretability of algorithms are particularly important.

2. Establishing a framework for algorithmic values and data ethics

When formulating regulations and policies related to algorithm governance, an ethical framework covering basic principles such as fairness, impartiality, transparency, and non-discrimination should be set up to motivate developers to take into account social responsibility and human well-being in algorithm design. At the same time, it is important to ensure that the security and privacy of personal data are not violated, and to prevent data abuse and leakage, so as to provide a guarantee for the healthy development of algorithms. By establishing clear ethical standards, the design and application of algorithms can be effectively guided in a beneficial direction, and this value and ethical framework not only protects the basic rights of users, but also promotes the social acceptance of algorithms.

3. Implement progressive regulation and multi-scenario evaluation

Excessive intervention or setting up too strict restrictions too early may hamper innovation or even lead to stagnation of technological development. In order to promote technological innovation while avoiding the inhibition of development caused by over-regulation, a progressive and phased regulatory model can be adopted. Instead of imposing across-the-board restrictions, moderate regulation should be imposed gradually according to the progress of technological development. For technologies that have been widely used and have far-reaching impacts, regulation should be increased; for technologies at the start-up stage, a certain amount of room for experimentation and innovation should be allowed, but basic bottom-line requirements should be set.

The risks and challenges faced by algorithms differ in different fields and application scenarios, so the regulatory framework should have flexibility and differentiated design. For example, in highly sensitive industries such as finance and healthcare, the regulation of algorithmic applications should be more stringent, focusing on issues such as transparency, interpretability, data privacy protection and avoidance of discrimination and bias of algorithms. For recommendation algorithms on entertainment or social platforms, on the other hand, the focus of regulation could be on preventing misleading information as well as guarding against over-reliance on algorithms. Flexibility in the design of regulatory frameworks can not only ensure the applicability of innovations across different industries, but also effectively manage potential social risks in key areas.

Algorithm governance has now become a global challenge. In the process of algorithm governance, faced with multiple challenges such as technological development, platform responsibility and public trust, the government, enterprises, academics and the public should join hands to adopt a comprehensive response strategy. By strengthening the construction of laws and regulations, promoting industry self-regulation, enhancing public participation and other aspects of the efforts to provide a solid foundation for the healthy development of algorithms. Only by ensuring that algorithmic technological innovation is accompanied by a proper response to potential social risks can we realize the harmonious coexistence of technology and society.