News

Exploring Decentralized Artificial Intelligence: Advancing the Democratization of GenAI

With the rapid development of Generative AI (GenAI) technologies—such as Large Language Models (LLMs), Multimodal Large Language Models (MLLMs), and Stable Diffusion—AI is increasingly permeating and transforming various industries, including life sciences, energy, finance, and entertainment. These technological breakthroughs not only accelerate innovation and enable personalized services but also significantly improve the efficiency of workflows. According to market forecasts, the global GenAI market is expected to grow from USD 40 billion in 2022 to USD 1.3 trillion over the next decade.   Challenges to Widespread Adoption and Strategic Countermeasures Despite its promise, the widespread adoption of GenAI faces substantial challenges. One of the most pressing issues is the concentration of GPU resources among major technology firms, which restricts the capabilities of research institutions and enterprises in developing their own models. Many organizations are forced to rely on API-based solutions, which not only introduce latency and security risks but also limit the customizability of models. Although open-source models offer some flexibility, they are often not sufficiently adaptable to domain-specific knowledge, hindering deep engagement by researchers in the pretraining phase, a critical stage for creating powerful and domain-aligned models.   In response, The Hong Kong Polytechnic University is pioneering an innovative GenAI infrastructure that enables enterprises and applications to independently pretrain their own GenAI models. This is achieved through a novel "Model over Models" (MoM) methodology to build foundation models. Specifically, global knowledge is divided into thousands of domains, with relatively lightweight Small Language Models (SLMs) trained for each. These smaller models demand far fewer resources—e.g., a 7-billion-parameter model can be continually pretrained using just 64 to 128 GPUs. Eventually, these SLMs can be integrated via the MoM framework to construct affordable and scalable Artificial General Intelligence (AGI) models, significantly lowering barriers to entry and enabling global participation in foundation model development.   For more details, please visit: https://www.stheadline.com/knowledge/3406043/

Prof. Yang Hongxia, Director of PAAI: " Aims to Drive World-Class AI Innovation in the Greater Bay Area”

On November 19, the Boao Forum for Asia Youth Forum 2024 Hong Kong Conference was held at the Hong Kong Convention and Exhibition Centre under the theme "Leading the Future: The Role and Contribution of Youth." Focusing on Artificial Intelligence (AI) and climate change, the event gathered over 30 experts and scholars from around the world. Through roundtable discussions and international youth dialogues, the forum aimed to advance Hong Kong’s AI industry and foster cross-border exchanges and collaboration among young people across Asia. At the event, Professor Yang Hongxia, a renowned AI scientist and faculty member at The Hong Kong Polytechnic University, was interviewed by Guangzhou Daily. Professor Yang highlighted the promising prospects and robust talent pool for AI development in Hong Kong. She expressed her hope to collaborate with institutions and enterprises in the Guangdong-Hong Kong-Macao Greater Bay Area to achieve world-leading technological breakthroughs in AI. Professor Yang previously served as Chief Data Scientist at Yahoo and has held senior research and leadership roles at IBM, Alibaba DAMO Academy, and ByteDance. She has authored over 100 publications in top-tier conferences and journals, and holds more than 50 patents in the U.S. and China. She joined PolyU in July this year. Commenting on her transition from industry to academia, Professor Yang candidly shared her motivation to address the barriers to entry in training large-scale AI models. “Training a large model often requires thousands of high-end GPUs over extended periods of time—resources that are far beyond the reach of universities and startups. Only tech giants have the capacity to do this,” she noted. Despite the fact that more than one million professionals work in the AI field globally, fewer than 1,000 have access to the core processes of large model development. “Technology evolution inevitably moves from centralization to decentralization—just like how early computers were massive machines, and now everyone can own one. This is the direction I hope to see,” she emphasized. Regarding her decision to relocate to Hong Kong, Professor Yang expressed strong confidence in the city’s future in AI development. She cited Hong Kong’s world-class education system, which nurtures high-caliber talent for the AI sector, as well as the government’s strong policy and financial support for AI initiatives. “Having recently participated in several project applications, I can say the support has been incredibly beneficial,” she remarked. She also praised policies such as the Top Talent Pass Scheme, which she believes will attract more top-tier AI professionals and scholars from around the world to work and conduct research in Hong Kong. “When talent and policy come together, it creates an ideal environment for AI development. This synergy is extremely valuable,” she added. During her time in industry, Professor Yang had already collaborated extensively with leading universities in mainland China. Since relocating to Hong Kong, she has continued to foster academic partnerships, including with Southern University of Science and Technology (SUSTech) and local institutions such as The University of Hong Kong and The Hong Kong University of Science and Technology. “Looking ahead, we hope to work closely with schools and enterprises across the Greater Bay Area to achieve world-class outcomes in artificial intelligence research and innovation,” she concluded.   For more details, please view via: https://huacheng.gz-cmc.com/pages/2024/11/20/86858f7b766c4daf9422b1bac1e954de.html