Google to Open AI Campus in Seoul This Year; Lee Says He Often Uses Gemini

By Kim Bongcheol Posted : April 27, 2026, 18:00 Updated : April 27, 2026, 18:00
President Lee Jae-myung shakes hands with Google DeepMind CEO Demis Hassabis at Cheong Wa Dae on the 27th. [Photo=Yonhap]
Google will open a Google AI campus in Seoul within this year to expand cooperation with researchers and startups. Google DeepMind and the government also agreed to build a cooperation framework for K-Moonshot, a government project aimed at AI-based innovation in science and technology.
 
Cheong Wa Dae said on the 27th that President Lee Jae-myung met with Demis Hassabis, Google DeepMind co-founder and CEO, often called the “father of AlphaGo,” and discussed ways to cooperate in the AI field.
 
The meeting was arranged as part of the government’s push to broaden global AI cooperation and work with top AI companies. Since taking office, Lee has met with OpenAI CEO Sam Altman, Nvidia CEO Jensen Huang and SoftBank Group Chairman Son Jeong-ui to continue AI-related cooperation, Cheong Wa Dae said.
 
The government also led the adoption of an “AI initiative” at last year’s APEC meeting in Gyeongju and is working with international organizations including the World Health Organization, the U.N. Development Programme and the International Telecommunication Union to establish a “global AI hub” in South Korea.
 
During the meeting at Cheong Wa Dae, Lee said, “I’m very interested in AI, and the country is investing a lot, but it’s hard to know whether it will go only in a direction that truly helps improve human welfare, or whether it could move toward attacking humans or harming world peace.”
 
Hassabis replied, “You raised a really important topic,” adding, “I think AI should be actively used to advance science and in the medical field. If it is used properly, it looks like it could bring major benefits to people around the world.”
 
Hassabis oversaw the 2016 match between Go champion Lee Sedol and the AI program AlphaGo. He also developed AlphaFold, an AI model for predicting protein structures, and won the Nobel Prize in Chemistry the year before last.
 
He said AlphaGo showed that his team could validate the technology and that it became a starting point for tackling harder problems. He said the goal was to expand those lessons into science and medicine, citing AlphaFold as a leading example that helps researchers better understand diseases.
 
Lee mentioned Google’s generative AI service Gemini, saying, “I use Gemini often, and sometimes it does things I didn’t ask it to do,” and asked whether that was “a kind of bug.”
 
Hassabis said foundation models can veer in a different direction if the guidance provided is not precise, and said safety measures known as “guardrails” must be built in when using and developing AI.
 
He added that as AI becomes more powerful, it will be given greater autonomy, sometimes called agent AI, and said safety controls will be essential if the world enters an era of artificial general intelligence, or AGI.




* This article has been translated by AI.

Copyright ⓒ Aju Press All rights reserved.