演講資訊
演講主題：AI 運算架構與演進趨勢 / AI Compute Architecture and Evolution Trends
演講時間及地點：2026/05/21星期四 14:00~16:00 臺大電機二館105演講廳
講者：梁伯嵩博士(聯發科技)
講者介紹：
梁伯嵩博士現為聯發科技企業策略與前瞻技術資深處長，並擔任聯發科技教育基金會董事。他同時兼任國立臺灣大學電機資訊學院資訊工程學系、電子工程學研究所及重點科技研究學院合聘之客座教授，與國立陽明交通大學電機學院客座教授。他同時也為 IEEE 中華民國分會理事，並於 2025 至 2026 年獲選為 IEEE 電路與系統學會產業傑出演講者。並擔任臺灣半導體產學研發聯盟常務理事，並曾於 2023 至 2024 年擔任 IEEE 電路與系統學會臺北支會主席。
梁博士於國立交通大學電子研究所取得博士學位，並於國立臺灣大學管理學院EMBA商學組取得商學碩士學位。其主要研究領域涵蓋 AI 運算架構、數位 IC 設計、處理器架構、量子運算及科技策略。他曾獲得多項重要獎項，包括中華民國十大傑出青年、經濟部智慧財產局國家發明創作獎的發明獎（共三次，包含一面金牌與兩面銀牌）、經濟部技術處產業科技發展獎之傑出青年創新獎、資訊月傑出資訊人才獎，以及由中華民國資訊學會與 ACM 臺北／臺灣分會頒發之李國鼎青年研究獎。
Dr. Bor-Sung Liang received the Ph.D. degree from the Institute of Electronics, National Chiao Tung University, and the master’s degree of business administration from the Executive MBA Program, College of Management, National Taiwan University. He is the Senior Director at Corporate Strategy and Strategic Technology, MediaTek Inc., Hsinchu, Taiwan, and the Director of the Board at MediaTek Foundation.
He is also concurrently serving as a Visiting Professor at the Department of Computer Science and Information Engineering (CSIE), the Graduate Institute of Electronics Engineering (GIEE), the College of Electrical Engineering and Computer Science (EECS), and the Graduate School of Advanced Technology (GSAT), National Taiwan University, and a Visiting Professor at the College of Electrical and Computer Engineering (ECE), National Yang Ming Chiao Tung University. He is a Director of the IEEE Taipei Section, and an Industrial Distinguished Lecturer of the IEEE Circuits and Systems Society (CASS) from 2025 to 2026. He is the executive director of Taiwan IC Industry & Academia Research Alliance (TIARA). He was the Chair of IEEE CASS Taipei Chapter from 2023 to 2024. His major research fields are AI computing architecture, digital IC design, processor architecture, quantum computing, and technology strategy.
He has received several important awards, such as the Ten Outstanding Young Persons, Taiwan, R.O.C., the National Invention and Creation Award on Invention (for three times, one Gold Medal and two Silver Medals) from Intellectual Property Bureau of the Ministry of Economic Affairs, Taiwan, the Outstanding Youth Innovation Award of Industrial Technology Development Award from the Department of Industrial Technology of the Ministry of Economic Affairs, Taiwan, the Outstanding ICT Elite Award of ICT Month, R.O.C., and the K. T. Li Young Researcher Award from the Institute of Information and Computing Machinery and ACM Taipei/Taiwan Chapter.
演講大綱：
人工智慧的發展重心已由學術研究轉向實際應用。然而，AI 發展在不同層面上仍面臨諸多挑戰。在本演講中，嘗試以結構化的方法，從多個不同觀點分析 AI 所面臨的機會與挑戰。我們提出一個人工智慧運算架構的七層模型，自下而上依序包含：實體層（Physical Layer）、連結層（Link Layer）、神經網路層（Neural Network Layer）、情境層（Context Layer）、代理層（Agent Layer）、協調層（Orchestrator Layer）以及應用層（Application Layer）。
我們亦運用所提出的七層模型，說明大型語言模型 (LLMs) 演進的三個階段。針對每一層，我們描述其發展軌跡與關鍵技術。在第 1 層與第 2 層中，我們討論 AI 運算所面臨的議題，以及向上擴充與向外擴展策略對運算架構的影響。在第 3 層中，我們探討 LLMs 的兩種不同發展路徑。在第 4 層中，我們討論情境記憶體對 LLMs 的影響，並將其與傳統處理器記憶體進行比較。在第 5 層至第 7 層中，我們討論 AI 代理（AI agents）與實體人工智慧（Physical AI）發展趨勢，並探討從單一 AI 代理演進至以 AI 為基礎的生態系過程中所面臨的議題，以及其對 AI 產業的影響。
The focus of AI development has shifted from academic research to practical applications. However, AI development faces numerous challenges at various levels. This talk will attempt to analyze the opportunities and challenges of AI from several different perspectives using a structured approach. We propose a seven-layer model for AI compute architecture, including Physical Layer, Link Layer, Neural Network Layer, Context Layer, Agent Layer, Orchestrator Layer, and Application Layer, from bottom to top.
It also explains the three stages in the evolution of large language models (LLMs) using the proposed 7-layer model. For each layer, we describe the development trajectory and key technologies. In Layers 1 and 2 we discuss AI computing issues and the impact of Scale-Up and Scale-Out strategies on computing architecture.
In Layer 3 we explore two different development paths for LLMs. In Layer 4 we discuss the impact of contextual memory on LLMs and compares it to traditional processor memory. In Layers 5 to 7 we discuss the trends of AI agents and explore the issues in evolution from a single AI agent to an AI-based ecosystem, and their impact on the AI industry.
主辦單位：教育部先進製程IC設計及驗證環境建置計畫
協辦單位：聯發科技-臺大創新研究中心/台大電子所/
報名連結:https://forms.gle/n7ncJnvkwKfNdzzi6
演講聯絡人:台大電子所專任助理 邱小姐02-33663718
