The alliance with the Service Science Society of Taiwan, held the “Generative AI Technology and Application” forum on the afternoon of August 9th at the Songshan Cultural and Creative Park in Taipei. The event had 38 in-person attendees and 12 participants online. A total of 11 short talks were arranged, featuring concise analyses by academic experts and industry professionals on the latest developments, concepts, and experiences in generative AI technology.
The forum began with an introduction by Professor Huang San-Yi, the project leader of the alliance and the president of the Service Science Society. He explained the background, noting that global investment in generative AI, particularly in large language models (LLMs), saw a tenfold increase in 2023. While the potential for development is highly anticipated, there has yet to be a significant impact on the daily lives and work of the general public. As a result, the industry is eager to explore its future commercial applications. Next, PhD student Tsai Yi-Hang from National Sun Yat-sen University introduced the latest developments in LLM technology, focusing on Retrieval-Augmented Generation (RAG), Multimodal LLMs, and Agentic Workflow. This was followed by the first session on “Generative AI Technology.”
The first short talk was given by PhD student Meng Chang-Chih from Professor Wu Yi-Cheng’s lab at National Yang Ming Chiao Tung University. He introduced the Llama TAIDE, a large language model developed under the National Science and Technology Council’s TAIDE project, which focuses on Taiwanese culture and Traditional Chinese, as well as the concept of fine-tuning multimodal models that integrate image recognition.The second short talk was presented by Jayron Yeh, Vice President of TWS Cloud, who introduced the Formosa LLM developed by TWS Cloud and the cloud and on-premises LLM services it offers.The third short talk in the first half was presented by Cheng Chi Chen, General Manager of Xuan Yi Technology. He introduced the LLM servers provided by the company, ranging from basic workstations equipped with 3090 GPUs for LLM inference to large servers paired with H100 GPUs for LLM training, covering computer systems for various application scenarios.The fourth short talk was given by Chun-Ying Lin, General Manager of Insynerger Technology, who introduced and demonstrated the company’s voice customer service solution. This service integrates ASR (Automatic Speech Recognition), TTS (Text to Speech), and LLM technologies to handle customer service tasks seamlessly across Mandarin, Taiwanese, and English in various industries.The fifth short talk was presented by Kai-Teng Chang, CEO of GenCode Technology, who introduced the company’s LLM multi-agent service construction system. This system allows for the creation of agents and workflows through a graphical drag-and-drop interface.
The second half of the forum focused on the “Applications of Generative AI.” The first presentation was by Yi-Chen Han, Vice President of Hamastar Technology, who introduced the application of generative AI in supply chains. She demonstrated how LLMs assist suppliers in completing sustainability questionnaires and how companies can use them to audit their suppliers.The second short talk was given by Ming-Hsin Liang, North Asia Director of iGroup, who discussed various applications of generative AI in education. These included automatic question generation, automated assessments, and the automatic summarization of research papers. The third short talk was presented by Chung-Hsuan Lee, General Manager of TPIsoftware, who introduced the application of generative AI in industry. He showcased the use of multimodal LLMs in semiconductor design and jewelry marketing, while also emphasizing the importance of business models. The fourth short talk in the second half was delivered by Yi-Hsun Ku, COO of QSearch, who introduced the application of generative AI in public opinion analysis. He explained how LLMs can automatically tag data scraped from the web in real-time for easier subsequent searches and better understand the sentiment of customer reviews to make more accurate recommendations. The fifth and final short talk was given by Wei-Liang Chen, General Manager of Moonstar Technology, who introduced the application of generative AI in project management. He demonstrated how LLMs can extract action items from meeting minutes and integrate them with project management systems for more precise tracking. Additionally, LLMs can be used to easily monitor project progress through a Q&A interface.
Through the insights shared by the 11 speakers, participants gained a deeper understanding of the latest trends and applications in generative AI technology. Please stay tuned for future events organized by the Text Mining and Digital Transformation Academia-Industry Alliance and the Taiwan Service Science Society.