Building secure, trusted AI: Insights from WIC Nishan Dialogue

By Ma Qing | (chinadaily.com.cn)| Updated : 2023-06-27

Print Print

23.jpg

Robin Li, co-founder, chairman and CEO of Baidu Inc, addresses the World Internet Conference Nishan Dialogue on Digital Civilization in Qufu, Shandong province, on June 26, 2023. [Photo provided to chinadaily.com.cn]

Robin Li, co-founder, chairman and CEO of Baidu Inc, said establishing comprehensive legal frameworks, institutional systems and ethical standards to ensure the healthy development of AI is vital to fostering a favorable innovation ecosystem.

He added it is important to prioritize risk prevention when developing AI by establishing mechanisms that allow for experimentation, adaptation and course correction to strike a dynamic balance between innovation and regulation.

As China has unique strengths in AI application scenarios, the development of specialized large models needs to be further tailored to sectors such as finance, healthcare and power.

By responsibly steering AI development and ensuring security measures, large models can shape a transformative digital landscape, he concluded.

24.jpg

Zhou Hongyi, founder of cybersecurity company 360 Security Group, addresses the World Internet Conference Nishan Dialogue on Digital Civilization. [Photo provided to chinadaily.com.cn]

Zhou Hongyi, founder of the cybersecurity company 360 Security Group, emphasized the future potential of large language models in enterprise, industry, government and city settings.

Large language models are a major advancement in intelligence and can significantly boost productivity. However, he pointed out limitations in current public models regarding industry specificity, data security risks, content trustworthiness and cost control.

He thus advocated for the development of vertical large-scale models tailored to specific domains, transforming their role from universal assistants to specific solutions for governments, industries and enterprises.

He added he believed it's important to create large-scale models that are secure, trustworthy, controllable and user-friendly. For example, when applied in enterprise settings, they should serve as assistants to businesses and employees.

< 1 2 3 >