Wenjun WU, State Key Laboratory of Software Development Environment, Beihang University
Tiejun HUANG, Academic Member of iGCU, School of Electronics Engineering and Computer Science, Peking University
Ke GONG, Chinese Institute of New Generation Artificial Intelligence Development Strategie, Nankai University
The rapid development and deployment of AI indicate an upcoming fundamental transformation of our society. This transformation can be a great opportunity to construct a human community with a shared future, and to promote the sustainable development of society and the natural environment. But without sufficient and effective governance and regulation, its implications might be unprecedented and negative. In order to ensure that these changes are beneficial before they are completely embedded into the infrastructure of our daily life, we need to build a solid and feasible AI governance framework to regulate the development of AI according to the ethics and values of humanity. In this way, we can make AI accountable and trustworthy, and foster the public’s trust toward AI technology and systems. To overcome the security problem of data-driven AI, research teams from companies and universities in China have endeavored to develop federated learning technology. To ensure the safety and robustness of DNN models, researchers have proposed new algorithms in adversarial testing and formal verification. Furthermore, research teams are investigating effective frameworks in the areas of AI interpretation, provenance, and forensics. These efforts are mostly in their preliminary stages and need further strengthening in order to deliver mature solutions for widespread adoption and practice. We suggest the following actions to push forward current initiatives on AI governance: Firstly, governments, foundations, and corporations should conduct cross-disciplinary, cross-sector, and multinational collaborations to establish a consensus on AI ethical principles. Secondly, they must intensify the collaborative research and development of AI governance technologies in order to keep pace with the rapid progress of AI. Thirdly, open AI DevOps platforms with built-in ethics-relevant tools should be developed to support all the stakeholders of different AI systems in evaluating the functional and regulation compliance of AI systems. Fourthly, clearly defined AI moral scenarios with significant social impact should be identified so that experts from different disciplines can work collaboratively to address the ethical challenges of AI. Lastly, we must actively promote ethical education for every stakeholder in AI research and development, application, and management, so as to significantly enhance their awareness of ethics and promote general practices of responsible conduct with AI.
AI ethical principles; AI governance technology; Machine learning; Privacy; Safety; Fairness
Wenjun Wu, Tiejun Huang, Ke Gong, Ethical Principles and Governance Technology Development of AI in China,Engineering, Volume 6, Issue 3, 2020, Pages 302-309, ISSN 2095-8099,https://doi.org/10.1016/j.eng.2019.12.015.