Vice Rector Zhou Wanlei shares insights on big data modelling and applications at Future Technology Salon

Vice Rector Zhou Wanlei of City University of Macau was invited to deliver a keynote speech titled “From Language Models to ChatGPT: Some Thoughts on the Applications of Big Data Models” at the 13th Future Technology Salon in the Zhuhai National Hi-tech Industrial Development Zone yesterday. He also participated in a roundtable discussion with experts and scholars about the impact of Artificial Intelligence (AI) on life and work.

[foogallery id="55925"]

The academic salon was a specialized session themed on AI and Big Data, supported by the Office of Network Security and Information Technology Committee of the Zhuhai Municipal Party Committee and Management Committee of the Zhuhai National Hi-tech Industrial Development Zone. It was organized by various departments of the hi-tech zone, including the Innovation and Entrepreneurship Service Center and the Technology Innovation and Industry Development Bureau. The event was attended by more than 100 people, including members of Zhuhai city and regional government departments, as well as representatives from the enterprise, university, and media sectors.

Professor Zhou Wanlei provided a thorough and comprehensive introduction to natural language processing (NLP), including the early development, the main technologies supporting ChatGPT, and the applications of ChatGPT. According to Professor Zhou, ChatGPT is currently most advanced language generation model developed by OpenAI, based on the transformer model. It can analyze a large amount of data and generate coherent and informative text. Professor Zhou Wanlei explained the operating mechanism of language models and related terminologies, including recurrent neural networks (RNN), long short-term memory (LSTM), and transformers.

Professor Zhou explained that the generative pre-trained transformer (GPT) models, including the recently launched ChatGPT-4, as well as its three previous generations, all employ transformers with self-attention mechanisms that enable the model to assign varying weights to different input and output sequences to infer meaning and context. ChatGPT has introduced a new approach that incorporates human feedback into the training process to better align the model’s output with user intent. Additionally, ChatGPT is equipped with a supervised fine-tuning (SFT) model, which enables it to reason in a similar way to a chain of thought.

Professor Zhou Wanlei emphasized that as ChatGPT matures in areas such as writing, drawing, modeling, video generation, and animation technology, it will bring revolutionary breakthroughs to existing software and applications in education, services, healthcare, consulting, entertainment, media, design, modeling, and other fields. He highlighted that “As a tool, it can help us develop new systems, optimize existing models, expand application scenarios, and improve work efficiency. At the same time, due to ChatGPT’s limitations, such as poor timeliness and potential for generating low-quality or harmful content, we need to be vigilant about the risks and challenges it may bring.” Finally, Professor Zhou Wanlei introduced the relevant research work and expressed the hope to establish long-term industry-academia collaborations with AI enterprises in the Zhuhai National Hi-tech Industrial Development Zone in the future, jointly promoting the development of the AI industry in Zhuhai and the high-tech zone.

Concurrently, Professor Zhou also participated in a roundtable discussion with experts and scholars, including Wang Dawei, Senior Operations Director of Kingsoft Office; Xu Shunda, Secretary-General of China University-Enterprise Committee of the China Association of Technology and Entrepreneurship; and Zhou Ze’an, Founder and CEO of Yoo Technology, to discuss the impact of AI on life and work.

[foogallery id="55934"]

Source: Zhuhai National Hi-tech Industrial Development Zone