|
An Interview with Mr. Tian-Tsair Su, President of Telecom Laboratories, Chunghwa Telecom Co., Ltd.
R&D Strategy for the AI 2.0 Era
Taiwan’s Chunghwa Telecom Laboratories, established in 1951, were formerly known as the Radio Research Agency, Directorate General of Telecommunications, and Institute of Communications Engineering, Ministry of Transportation and Communications. When Chunghwa Telecom started privatization and transformation in 1996, the Laboratories were re-structured and eventually became the “Chunghwa Telecom Laboratories” in 2012. The Laboratories have been dedicated to developing telecom technologies for a long time, have been involved in various applications of information and communication technology(ICT), and have advanced research programs in many. As a renowned R&D institute in Taiwan's information and communication sector, the Laboratories actively engage in major international standard organizations and collaborate with prominent global think tanks.
Growth and Research Focus The Telecom Laboratories have seen continuous growth in its R&D efforts, expanding from one research unit in 1951 to nine units with approximately 1,200 personnel today, marking a 40-fold increase in organizational scale. Over 96% of its staff hold advanced degrees, including master’s and doctoral degrees. The institute’s R&D areas can be broadly categorized into eight fields: fixed-line broadband, mobile broadband, operations management, cloud computing, digital innovation, smart networking, artificial intelligence (AI), and cybersecurity. Expansion into AI 2.0 Era Many wonder if a large-scale research organization like the Telecom Laboratories will continue to invest in AI and optical communications/silicon photonics technologies, and how it will harness these key technologies to foster the growth of the domestic industry. Mr. Tian-Tsair Su, president of Telecommunication Laboratories, explained that the institute’s R&D focus is divided into two main areas: basic ICT infrastructure and innovative applications. The topics in basic ICT infrastructure includes fiber optic networks, wireless networks, satellite communications, network management, and cloud computing, while the topics in innovative applications cover AI/big data, IoT, cybersecurity, blockchain, and the metaverse. The institute has always placed a strong commitment to research in AI and optical communications. For example, since 2016, it has actively adopted AI technologies in speech recognition, image recognition, text analysis, and network operations. Achievements include text-based customer service chatbots, outbound call bots, voice-controlled MOD service assistants, customer sentiment analysis solutions, facial and body recognition solutions, intelligent enforcement image recognition solutions, and intelligent network maintenance (such as quality degradation analysis and recovery). In November 2022, with the rise of generative AI technologies, such as image generation, conversational AI, speech synthesis, and text/music generation, the institute swiftly allocated resources to develop diverse AI 2.0 consumer and enterprise applications. These applications include intelligent customer service knowledge Q&A Copilot, smart medical AI assistants, intelligent judicial and legal models, smart traffic management, intelligent film and TV chatbots, real-time multilingual translation and voice imitation robots, SMS/voice fraud detection and prevention applications, and AI audiobooks. Additionally, the institute is focusing on developing AI 2.0 computing cloud resource management systems and AI 2.0 application development platforms to efficiently provide diverse AI cloud services, aiming to enhance AI training speed by 40%, GPU computing efficiency by 3 times, and AI 2.0 application development efficiency by 2.5 times. Looking ahead, the institute will continue to develop diverse, ultra-fast, and low-latency AI applications to meet various user needs. Innovative Achievements and Collaborations President Su highlighted the institute’s significant innovative achievements, noting that its technical innovations are highly valuable. For instance, in 2023, the institute’s innovative patents were showcased at the Innovation and Invention Exhibition, where it won the most awards in the ICT industry. Most of these innovations are software-related, and there is a strong desire to collaborate with domestic hardware manufacturers for integrated solutions to enhance the commercialization of research outcomes. Moreover, the institute has actively engaged in international standard organizations. Starting in 2023, it partnered with renowned international telecom companies such as NTT, Orange, TIM, and Verizon to establish the “Autonomous Networks Hyperloops - Virtual Command Center as a Service” Catalyst Project. This project integrates key technologies such as autonomous networks, AI, and digital twins to build a Virtual Command Center (VCC) that helps enterprises strengthen disaster response measures and ensure the continuous operation of critical infrastructure. In June 2024, the institute demonstrated its forward-looking research results at the TM Forum annual conference, receiving the TM Forum Outstanding Catalyst Award - Beyond Telco. Focus on All-Optical Cloud Network Development Given that all-optical networks represent a significant global trend in network evolution, how does Chunghwa Telecom plan to establish and leverage the advantages of all-optical network technologies to achieve high capacity, low latency, and low power consumption? President Su noted that since 2020, Chunghwa Telecom has joined the IOWN (Innovative Optical and Wireless Network) Global Forum, engaging in technical exchanges with NTT, Fujitsu, the Industrial Technology Research Institute (ITRI), and domestic network communication vendors. The company has successfully created a long-distance (380 kilometers), low-latency (approximately several dozen milliseconds), and high-interactivity all-optical holographic exhibition application verification field at its Taoyuan and Kaohsiung data centers. Plans are underway for a cross-national demonstration with NTT in November this year, and the technology will be showcased at the 2025 World Expo in Osaka, Japan. President Su also mentioned that the institute encourages its R&D teams to pursue startup opportunities and spin off into subsidiaries. Although internal entrepreneurship faces challenges, the institute strongly supports innovative ideas from its teams. For example, since 2021, Chunghwa Telecom has established a "Startup Incubation Center" within the institute, providing annual funding of up to NT$100 million to startup teams, along with competitive salaries. The incubation center actively connects teams with external coaching resources and integrates company resources to accelerate market development, providing significant support for the teams. Pursuing Excellence in Innovative R&D Regarding the development and training of AI talent and the integration of AI with optoelectronics, President Su emphasized that the growth of various AI applications is closely linked to high-quality, high-bandwidth networks. This means that networks will accelerate their evolution towards all-optical and silicon photonics architectures to foster more diverse AI applications. In addition, Chunghwa Telecom Laboratories is actively working on integrating these AI applications with its all-optical cloud network technologies. Since 2016, the company has been planning and promoting AI talent development, including arranging a series of AI courses at Chunghwa Telecom Academy to help all employees enhance their AI skills. Additionally, the company continues to recruit experienced AI professionals through its website and external recruitment channels. For a long time, the Telecom Laboratories has been known for cultivating outstanding talent, establishing a prestigious reputation, standing as a leading force in telecommunications research. Building on the solid foundation laid by pioneers in telecommunications, the institute is committed to its vision of “innovative R&D, pursuing excellence, and becoming a world-leading research organization.” |