Building a Secure and Trusted Artificial Intelligence, Focusing on the Principles of Digital Civilization Development | Models | Focus
In today's world, digital technology is fully integrated into various fields and processes of human society, and artificial intelligence has become an important symbol of human civilization progress. At the same time, if digital technology loses its norms and constraints, it may become a "Pandora's Box" and pose a certain threat to the real society. On June 26, the main forum of Nishan Dialogue on Digital Civilization of the World Internet Conference was held in Qufu, Jining, Shandong Province. During the conference, Chinese and foreign guests made suggestions on the topic of "building safe and reliable artificial intelligence".
Artificial intelligence will reshape the digital world
"The opportunities for artificial intelligence are enormous. However, with the accelerated global application of artificial intelligence, it is necessary for governments and organizations to consider how to best utilize this technology to benefit humanity and the earth." Hong Yaozhuang, CEO of the Global Association of Mobile Communications Systems, said that globally, artificial intelligence is gradually emerging as a powerful force with the potential to change businesses and society.
Mohammad bin Amo, Secretary General of the Arab Information and Communication Technology Organization, stated that artificial intelligence has revolutionary potential to influence and even change all aspects of life, drive innovation, and make life more convenient. For example, improving efficiency through automation, providing accurate information to support informed decision-making, and improving customer service and experience.
When it comes to the rapid development of AI, Robin Lee, the founder, chairman and CEO of Baidu, also has deep feelings. He said that in the past year, AI has advanced at an iterative speed of "week" in all aspects of technology, products and applications. The big model will penetrate into more and more fields, and the deep integration of the digital economy and the real economy driven by the big model will strengthen, optimize, and expand the real economy, create considerable incremental value, and bring about profound changes in economic and social development and industries.
"The birth of the GPT big model represents the arrival of universal artificial intelligence and strong artificial intelligence, and is a true emergence of intelligence." Zhou Hongyi, founder of 360 Group, believes that the AI big model solves all natural language processing problems with a set of models and algorithms, has evolved from perception to cognition, can understand text and language, make analysis and planning, and traditional algorithms will be replaced. In many important fields such as autonomous driving and robot control in the future, big language models will demonstrate their skills.
Zhang Yong, Chairman and CEO of Alibaba Group and Chairman and CEO of Alibaba Cloud Intelligence Group, stated that in recent times, global technology companies have been competing for universal big models, driving artificial intelligence to a new stage of development. He believes that big models are accelerating the arrival of the era of intelligence, and the real economy is the main battlefield of intelligence.
"We are witnessing the accelerated development of artificial intelligence," said Bart Selman, former chairman of the Association for the Advancement of Artificial Intelligence and professor at Cornell University. GPT-4 has achieved the first form of universal artificial intelligence, introducing a new entity that will inevitably change many existing social structures and mechanisms, and humanity needs to rethink the impact that artificial intelligence will bring.
The governance challenges brought by artificial intelligence cannot be ignored
"We urgently need to develop and apply artificial intelligence with a high sense of responsibility to ensure that it has a positive impact on society as a whole." Mohammed Ben Amo said that while enjoying the benefits of artificial intelligence, it is necessary to recognize the many social and ethical challenges it brings, which need to be addressed from the beginning.
The artificial intelligence big model not only brings universal development opportunities for human society, but also brings many governance challenges that cannot be ignored. Robin Lee believes that the application of new technologies often precedes the norms. Only by establishing and improving laws and regulations, institutional systems, ethics and morality to ensure the healthy development of AI, can a good innovation ecology be created. Looking towards the future, while paying attention to risk prevention, we should also simultaneously establish fault-tolerant and error correction mechanisms, and strive to achieve a dynamic balance between standardization and development.
In Zhou Hongyi's view, the public ownership model has many shortcomings, such as a lack of industry depth, susceptibility to internal data leakage, inability to ensure the authenticity of content, and inability to achieve cost control for enterprises. He believes that the application of large models in enterprise level scenarios should adhere to the "auxiliary mode". The key to achieving "safe and controllable" GPT model lies in whether the enterprise can truly control it, and let people's will play a crucial role in the entire decision-making loop.
Currently, the global competition in artificial intelligence has entered a white hot stage. In this situation, Zhang Yong said that everyone hopes to make a difference, but the technical and ethical issues in the relevant field still need to be clarified. It is also important to wear safe brakes while stepping on the accelerator. "The evolution of AI technology must conform to social norms and bring more beauty to mankind. While pursuing progressiveness technology, we must adhere to a sense of social responsibility."
Building a secure and trustworthy artificial intelligence has gradually become an industry consensus
Walter Fome, Chairman of the Joint Technical Committee on Network Security and Data Protection of the European Commission for Standardization and the European Electrotechnical Standardization Committee, stated that in order for artificial intelligence standardization to succeed, a common definition of artificial intelligence vocabulary among countries is needed, which will promote a common understanding of secure and trustworthy artificial intelligence.
The enterprise level scenario requires the creation of a proprietary large model that is safe, trustworthy, controllable, and easy to use. So how to create such a proprietary big model? Zhou Hongyi stated that several principles need to be adhered to: security principle, enterprise level applications must be secure big models; The principle of trustworthiness is used to address the accuracy and real-time shortcomings of large models; The principle of controllability should position the large model as an auxiliary tool to ensure that people are in the loop; The principle of usability requires a "soulful digital assistant" to solve usability problems; The principle of inclusiveness makes big models a productivity tool that everyone can use.
There are many uncertainties in innovation, some of which can be predicted in advance and prevented in advance; Some problems arise during development and need to be solved while developing, using development to solve them. Zhang Yong stated that "building a secure and trustworthy artificial intelligence" has gradually become an industry consensus, and relevant laws and regulations have been gradually improved, cultivating a good soil and environment for the sustainable development of technology and industry.
AI will lead a new wave of technological innovation. Digital technologies such as 5G mobile communication, industrial Internet, big data and AI have changed the way people produce and live. Liu Liehong, Chairman of China United Network Communication Group Co., Ltd., stated that the goal of the whole society is to enable everyone to share the dividends and achievements of digital civilization development.
Mohammad bin Amo believes that the future of artificial intelligence is full of hope, and a secure and trustworthy artificial intelligence should be built around five aspects: privacy, transparency, responsibility and accountability, robustness, and resilience. In addition, an ethical framework should be established and human welfare should be prioritized through a people-oriented design, contributing to the sustainable development of building safe and trustworthy artificial intelligence.