Signing Ceremony of ZGC Global High-Level Think Tank Alliance, Beijing, April 28./CFP
Signing Ceremony of ZGC Global High-Level Think Tank Alliance, Beijing, April 28./CFP
Editor’s note: The article is written by Keke Gai, professor at the School of Cyberspace and Science and Technology, Beijing Institute of Technology, and Jing Yu, associate professor, the Institute of Information Engineering, Chinese Academy of Sciences. The article reflects the authors’ opinions and not necessarily the views of CGTN.
The 2024 Zhongguancun Forum is held in Beijing from April 25 to 29, with the theme “Innovating for a Better World.”
Early this year, China announced a “new quality productive forces” concept in order to promote high-quality development, considering both green and digital innovations. The concept of the new quality productive forces differs from traditional economic growth methods and production development paths, which align high-speed development with a number of dimensions, such as high technology, high efficiency, and high quality.
In the 2024 Zhongguancun Forum, AI (Artificial Intelligence) is undoubtedly the star of the show, attracting many practitioners in the industry/academy to participate and share their insights. It is easy to observe from diverse range of AI products available that AI technologies provide immeasurable opportunities for empowering the development of new quality productive forces.
First of all, the general AI industry will be powered up by policies, according to the forum. An official document suggests a number of measures for accelerating the development of the general AI industry, which covers computing power, infrastructure, data-driven applications, large AI models, ecosystems, etc. In order to encourage innovations, various methods will be applied, such as “unveiling the list” and “horse racing.” These supports will be beneficial for making breakthroughs in the AI field.
Next, AI technology will revolutionize a wide variety of industries. For example, AIGC (Artificial Intelligence Generated Content), as an emerging technology, is reshaping various domains with exciting opportunities. It has been widely believed that AIGC will bring a revolution in healthcare, as the technology is able to enhance efficiency, cut down costs, and improve treatments, e.g., diagnosis, medicine development, and telemedicine. Financial services are another beneficiary of adopting AIGC, because AIGC shows great potential in optimizing service processes and strengthening decision-making.
Finally, more opportunities in education will be made by AI adoptions. Since AIGC brings a new way of interactions between humans and machines, it makes a personalized learning experience possible. With the introduction to AIGC-enabled interactive platforms, we can predict that future education will become more flexible, attractive, and planned for learners, which also implies new and emerging opportunities for many.
2024 Zhongguancun Forum on Industry-Scale Model Innovation and Development on April 28 at Beijing./CFP
2024 Zhongguancun Forum on Industry-Scale Model Innovation and Development on April 28 at Beijing./CFP
Although AI technology offers significant powers for the new quality productive forces, we still need to pay attention to the challenges. From the perspective of the technology, i.e., AI, an area of significant challenge lies in cybersecurity.
First, AI security is encountering threats to endogenous security and how to construct trustworthy AI solutions is a challenge. For example, in AIGC, how to ensure the security of model corpus requires consideration of various aspects, e.g., source security, annotation security, and adversarial attacks. In terms of securing large AI model services, there are many unsolved issues, such as guaranteeing service transparency, data manipulation, and counterattacks.
In addition, exogenous safety is another challenge that determines whether AI technologies are governable. Model content security is a concern that has not yet been solved, covering content generations’ accuracy, reliability, and security. In order to achieve secure data commercialization, model identity and verification technologies is a basis; however, practical solutions are being explored. Meanwhile, model safety assessment technology also requires a lot of explorations.
Finally, AI-derived security issues present challenges in achieving controllable AI. In terms of secure model usage, more work is required in multiple dimensions, such as fundamental theory of model usage and the cognitive safety of using AI. Even considering the existing AI services, we still deal with various technical obstacles, for example, data isolation issues in healthcare. Contemporarily, R&D work has made efforts to address challenges in this domain, such as using blockchain technology and privacy computation to solve security and privacy issues and facilitate data commercialization.
In summary, we can see that AI technology security is a vital power that contributes to new quality productive forces. Addressing and solving issues in AI, especially security issues, is a key in determining how much value AI technology can actually contribute.