It will take years or decades for the development of artificial intelligence (AI) to reach human-level AI. Robots will not take over the world. “Open source” is the only option.
These are among the thought-provoking insights that Turing Prize Winner Prof. Yann LeCun, hailed by the media as one of the “Godfathers of AI,” has shared with AI experts, scholars, industry leaders from around the world and students at a symposium held recently in Hong Kong.
Entitled AI NEW HORIZONS 2023: A Symposium with Scientific Leaders, the symposium was jointly hosted by the Hong Kong University of Science and Technology (HKUST) and the Greater Bay Area Association of Academicians (GBAAA).
In his keynote speech, Prof. LeCun envisioned objective-driven AI will help predict the future, and Joint Embedding Predictive Architecture (JEPA) model will make a paradigm shift in predictive modeling, that JEPA could bring about a “new Renaissance.”
AI, Prof. LeCun said, will become a “shared infrastructure” in the future, like the Internet today. That also means, he said, AI platforms must be made “open source.” “All interactions with the digital world will be mediated by AI assistants… They will constitute a repository of all human knowledge and culture,” the leading AI expert said.
The issue of “open source” AI platform was further discussed at a “Fireside Chat” among Prof. LeCun, HKUST Council Chairman and AI expert Prof. Harry Shum and Director of HKUST’s Center for AI Research Prof. Pascal Fung, at the symposium held at the Asia Society Hong Kong Center.
Admitting that the issue was complex and controversial, Prof. LeCun said open source platform would promote diversity. Prof. Shum shared the view that “open source is a good thing,” but added the AI industry would be divided on the idea in view of their different considerations in developing Large Language Models (LLM).
“If you think from a commercial point of view, whoever is Number One (in the industry) would not want to go open source for financial reasons. If you are Number Two, you will then struggle (and ask) should I open source or compete with (Number One)?” Prof Shum said.
“Like many new technologies, it would take time to formulate the relevant laws and regulatory mechanisms to tackle the problems that may arise,” he added.
Responding to the question by Prof. Fung about universities and industry co-operations, both Prof. LeCun and Prof. Shum underlined the importance of greater collaborations between the industry and the academia in AI. Prof. Shum said resources were important to research development, “we will continue to encourage faculties and students to get more fundings and do start-ups… This is what HKUST has been doing.”
Prof. Nancy Ip, HKUST President and GBAAA Council Chair who spearheaded the symposium, expressed her deep appreciation to the experts and scholars who shared their valuable insights at the event. The symposium, she said, was aimed at “exploring the current AI research landscape and delving into the cutting-edge work being undertaken, to understand the potential AI holds as well as its limitations.”
“The launch of Open AI’s ChatGPT and other generative AI tools early this year has suddenly thrusted AI into the forefront of our life and imagination,” she said. “As AI research progresses, we can anticipate even more transformative advancements that will redefine the way we work and live.”
As an officiating guest of the event, Secretary for Innovation, Technology and Industry Prof. Sun Dong, said Hong Kong possesses distinct competitive advantages to be developed into an international I&T center. He outlined the Government’s I&T strategy and plans to digitalize public services set out in the 2023 Policy Address, which include the establishment of a supercomputer center to foster AI development.
The symposium was the first large-scale academic conference held by the GBAAA since its establishment in 2021, and also one of the series of conferences recently organized by HKUST with a focus on AI.
Scholars from different regions and senior executives from leading innovation technology firms attended the event. They include Prof. Kathleen McKeown at Columbia University; Dr. FENG Junlan at China Mobile; Dr. LI Hang at ByteDance; and Dr. ZHOU Jingren at Alibaba Cloud Intelligence.
Representatives from the Chinese Academy of Sciences and eight local tertiary institutions including the City University of Hong Kong, Hong Kong Baptist University, Lingnan University, The Chinese University of Hong Kong, The Education University of Hong Kong, The Hong Kong Polytechnic University, HKUST and The University of Hong Kong, took part in the full-day symposium and shared with participants their AI research achievements.