Recently, an open letter calling for a temporary halt to the research of AI models more advanced than GPT-4 has sparked widespread discussions in the global tech industry. In this context, our interviewed Professor Cai Hengjin from the School of Computer Science at Wuhan University to understand his views on the matter.
Professor Cai Hengjin said that the rapid development of AI technology is inevitable but needs more attention and scrutiny to ensure the safe, reasonable, and ethical development of AI. He noted that some existing AI technologies, such as ChatGPT, have indeed shown astonishing development speed and capabilities, but there are still many issues to explore and resolve concerning the relationship between AI and humans.
Professor Cai also emphasized that the development of AI technology requires broad social participation and cooperation, rather than being decided by a few companies or individuals. He called on governments and academic institutions to strengthen regulation and research, and establish stricter standards to ensure the safe development of AI technology.
Professor Cai Hengjin elaborated on the proposed graded supervision and governance of AI: the first level is treating AI as a tool, the second level is treating AI as an avatar of human subjects, and the third level is treating the Metaverse as a venue for the realization of superintelligence.
Treating AI as a common industrial product or tool, we need to consider the responsibility of producers, consumers, and the inherent risks of AI. These responsibilities and risks can be transferred to professional institutions through market mechanisms. Fully autonomous vehicles are transportation tools, and the responsibility for accidents can be bundled with insurance companies.
AI as an avatar of a human subject requires the subject to assume vicarious liability. Legitimate AI avatars need to pass certain tests and pay a fee to obtain quasi-legal person status. For malicious or erroneous avatars, the ultimate punishment can be the deprivation of their quasi-legal person status.
The Universe corresponds to the material world, while the Metaverse corresponds to the world of consciousness above the material world, serving as an extension of human consciousness or its externalization and objectification in digital space. The Metaverse can serve as an entry point for humans into the AI world, with our avatars acting as nodes in the Metaverse, coexisting and thriving with other computing, storage, and sensing nodes and evolving together.
The Metaverse will evolve into a "super-brain" in near-Earth space, representing superintelligence in the human world.
Regarding the future relationship between AI technology and humans, Professor Cai Hengjin stressed that AI is not the enemy of humanity. It can become a partner, making even greater contributions to human development and well-being. He believes that the development of AI technology must follow certain ethical guidelines and shared human values, and it requires broader social participation and joint efforts to realize its true value and potential.
At the end of the interview, Professor Cai Hengjin expressed his confidence in the future of AI technology, while also calling for collective efforts to explore and innovate the development path of AI technology, enabling it to better serve the social and economic development of humanity.
Expert Profile: Cai Hengjin, Professor at the School of Computer Science at Wuhan University, expert on the Science Committee of the Asia Academy of Digital Economics . The Asia Academy of Digital Economics is headquartered in Singapore and consists of outstanding scholars and scientists in the digital economy field, committed to promoting the vital role of science and technology in enhancing social productivity and regional economic development through the exchange of cutting-edge research results and advanced technology development.