Global Scholar Yann LeCun: "Open Source Platform Needed for AI at Human-Level Performance"

Yan LeCun, a world-renowned artificial intelligence (AI) scholar and professor at New York University, stated that an open-source platform is necessary for developing AI that can be utilized in real life.

On the morning of the 11th, Yann LeCun, Meta's Chief AI Scientist and a professor at New York University, delivered the keynote speech at the "2024 K-Science & Technology Global Forum" held at the Fairmont Ambassador Hotel in Yeouido, Seoul. Photo by Seoyul Hwang chestnut@

원본보기 아이콘

In his keynote speech at the '2024 K-Science & Technology Global Forum' held on the morning of the 11th at the Fairmont Ambassador Hotel in Yeouido, Seoul, Yan LeCun, Meta's Chief AI Scientist and NYU professor, said, "It is foolish to block open-source AI platforms through regulation."

He elaborated, "Resources should be allocated so that computing resources are used appropriately, and data reflecting the languages and cultures of each country should be accumulated as much as possible to enable AI development." Professor LeCun delivered the keynote speech on the theme of 'Advancement of AI Technology and Global Cooperation.'

The reason for this suggestion is that, in LeCun's view, AI technology trained based on language has not yet developed to the level humans desire. He said, "Large Language Models (LLMs) do not fully reflect the complexity of the real world," and "Since LLM training is still insufficient, a practical AI revolution has not yet occurred."

He then proposed the 'Zephyr' model, which learns not only language but also video data. Professor LeCun explained, "It is about training AI to predict all the abstract things that happen in videos," and added, "I understand that similar architectures are being researched in Korean academia." He further mentioned that opportunities for collaborative research on this will be provided through the Korea-US joint AI research center, the 'Global Frontier Lab,' established in New York.

LeCun emphasized the open-source platform, stating that all human knowledge should be accumulated through AI based on the Zephyr architecture.

He said, "Language should be accumulated through AI, and for this, companies and institutions from each country must collaborate and share data based on open source," adding, "India alone has 7,000 dialects, and Vietnam also has multiple languages. Data accumulated from just one region in the western United States cannot create AI that humans can use in real life."

Professor LeCun cited the internet as an example of open-source-based technology. He explained that all elements used for communication on the internet, such as protocols and software, are open source. He said, "My hope is that public, private, cloud service providers, and large companies like Meta will collaborate globally to create foundation models," emphasizing, "When AI is created with the accumulation of cultures, languages, and knowledge from around the world, it will be AI that we can use."

Meanwhile, Professor Eunmi Chae of Korea University’s Department of Physics, who gave a keynote speech on the theme of 'Growth of Young Scientists through International Scientific and Technological Cooperation,' shared her experience of joint research in the United States and other countries, emphasizing, "Exchange among researchers who grew up in different cultural backgrounds brings about significant changes in science and technology overall." Professor Chae is an emerging researcher in quantum computing, having earned her bachelor's and master's degrees from the University of Tokyo in Japan and her Ph.D. from Harvard University in the U.S., where she conducted research on laser cooling of diatomic molecules.

© The Asia Business Daily(www.asiae.co.kr). All rights reserved.