While deeply integrating with the real economy and social life, AI will pose new challenges for the existing production relations, social governance and other aspects, and also raises some thoughts and concerns about the ethical side of AI. It's a significant task posed before us as to how to precisely grasp the features of the era, propose reasonable and forward-looking ethic rules, construct an international framework system for AI governance and cultivate a new ethical order that's more just and humanized.

On November 10, at the Artificial Intelligence and Digital Ethics Forum, the 2022 World Internet Conference, ZHU Shiqiang, President of Zhejiang Lab, delivered a speech on the "Technical Logic of AI Ethics and Human's Options".

"Weak AI is strong and strong AI is weak. These are the two features of the current development of AI." ZHU Shiqiang said that AI currently can match or outreach humans in certain aspects, such as AlphaGo defeating human masters at Go or AlphaFold predicting the structure of a protein. But in the meantime, AI lacks the human ability to learn, think and feel.

"AI is just the container of human wisdom. We shall take control all the time. The human can create sophisticated intelligent agents and thus can also control them. Our priority at current stage is still accelerating technological development while remaining alert to possible ethical issues and taking precautions accordingly", said ZHU Shiqiang.

To cope with possible ethical issues caused by AI in the future, Zhejiang Lab has formed a set of AI ethics diagrams which, from the perspectives of R&D, use and management, defined the possible problems faced by different parties and their respective responsibilities. From the perspective of R&D, scientific researchers need to overcome technical issues such as data prejudice, data security and algorithm explainability; from the perspective of use, certain important functions and tools need to be developed in order to avoid the digital gap, the Matthew Effect, social prejudice, etc.; from the perspective of management, legal means need to be implemented to prevent misuse of AI, including leakage of private information and sophisticated counterfeit.

"We need to use technology to restrict AI and make AI knowable and controllable, secure and trustworthy, fair and just, and agile in governance, so as to achieve the purpose of AI benefiting mankind", said ZHU Shiqiang. Based on the ethics diagrams, the Lab has taken actions from various aspects to prevent the possible social harm caused by AI.

First, define technological direction by technical standards. The Lab has been approved for the establishment of a National Technical Standard Innovation Base (Intelligent Computing), and has led or participated in the formulation of 22 international and national standards related to intelligent computing, including leading the formulation and release of the first international standard on "Technical Framework for Shared Machine Learning System".

In the meantime, focusing on the new generation of AI, the Lab, by constructing a professional and comprehensive knowledge base, is to build the dual engine of data and knowledge to drive AI development, and establish the new generation of explainable methodology system to support the formation of transparent intelligence. It developed the "ZJ Dubhe" AI open source platform, post-quantum password key technology and platform, data privacy computing platform and other tools to prevent malicious applications.

Besides, the Lab is researching on a full-dimensional definable network that can evolve its infrastructure and is flexible, definable and reconfigurable, and building comprehensive experimental facilities for national industrial control system security with the functions of test verification, digital twin and monitoring and evaluation, so as to make breakthroughs in key technology and equipment and ensure network security.

"We need to be more patient to support the development of technology and remain highly attentive to the possible risks caused by AI. We hope that with technological development and based on legal norms, the digital civilization that's more just and reasonable, open and inclusive, secure and stable and vibrant, will flourish in China", said ZHU Shiqiang.