Trustworthy artificial intelligence technologies and the reliability of large language models have been in the spotlight at the World Artificial Intelligence Conference 2023.

Joint efforts have been put forward by academics, researchers and government authorities to call for reliable and ethically responsible artificial intelligence solutions.

An LLM is one of the hottest types of AI algorithm that uses deep-learning tools and large data sets to understand, summarize, generate and predict new content.

At the WAIC, the China Academy of Information and Communications Technology initiated the Large Language Model Ecosystem Collaborative Community which brings together more than a dozen leading players such as Alibaba, Baidu, JD and Cloudwalk.

It looks to focus on key tasks such as setting up open service platform of LLMs, industry-academia-research exchange, as well as mitigating potential governance risks.

Local digital players at the WAIC have released their latest offerings, ranging from image detection tools to AI content assessment frameworks.

The Ant AI Security Detection Platform unveiled new functions to offer artificial intelligence generated content (AIGC) assessment tools, as the authenticity of digital content has been a rising concern. In addition, it also introduced the "zero-knowledge proof" computing technology to its blockchain architecture to help verify information without leaking the content.

"To build trustworthy AI solutions, various factors including data security, system traceability and accountability, network environment credibility and legal ethics shall all be taken into account," commented He Jifeng.

He is an academia of the Chinese Academy of Sciences and director of the Shanghai Academy of Artificial Intelligence Industrial Technology.

A proposal has been jointly put forward by the Shanghai Artificial Intelligence Lab, China Academy of Information and Communications Technology, Wuhan University and Ant Group to promote trustworthy AI capabilities and push forward the research for relevant solutions.

It also hopes to enhance the awareness for key issues such as data governance, privacy and ethics.

Experts also shared the latest development to assess numerous digital content due to the increasing amount of generative AI content.

Tang Qi, general manager of the Smart Intelligence Unit from Shanghai-based IntSig Information, explained the crucial technology advancement for image tampering detection technology to recognize fake or falsified photos, smartphone screenshots or even bank statements.

Such capabilities have become crucial to safeguard the authenticity of online content and sensitive information, he added.

On Saturday, a trusted data economy demonstration zone was unveiled in Jing'an District to showcase the city's continuous effort to integrate private and public segments to enhance the digital economy.

Geoff Jiang, senior vice president of Ant Group, said multiple technologies including the blockchain, Internet of things and privacy computing would offer more valuable solutions for different industries.

Tencent has upgraded its Model-as-a-Service platform to allow easier deployment and accessible LLMs for different scenarios, covering more than 50 specific industries.

"There are still challenges to build industry specific LLMs, such as limited computing power, poor data quality, high investment cost and security and compliance concerns," Tencent's vice president and head of the Youtu AI Lab Wu Yunsheng said.