Stanford AI Index 2024 Report
South Korea's AI Investment Halved Last Year
Concerns Rise Over Weakened AI Development Competitiveness
Despite South Korea possessing promising talent in the field of artificial intelligence (AI) technology, it has been experiencing low development performance and talent outflow issues.
According to the 'AI Index 2024' report published on the 15th (local time) by the Human-Centered AI Institute (HAI) at Stanford University, the number of AI patents per 100,000 people in South Korea in 2022 was 10.26, ranking first among the surveyed countries. This figure significantly surpassed the 3rd place United States (4.23) and 4th place Japan (2.53), as well as exceeding the 2nd place Luxembourg (8.73) by 1.5 patents.
However, there were no development achievements last year in 'foundation models,' which are the basis of generative AI technology. The United States developed the most with 109 models, followed by China and the United Kingdom with 20 and 8 models respectively. The United Arab Emirates (UAE) also created 4 models. Even Egypt had 2 models listed in HAI's 'notable AI models' list, but South Korea had none.
Furthermore, the issue of talent outflow was revealed. The AI talent outflow index per 10,000 Korean members registered on the business networking platform LinkedIn recorded -0.3. This means that AI talent is leaving South Korea rather than entering it. The talent mobility index was highest in Luxembourg (3.67) and Switzerland (1.60). The United States recorded 0.40.
This poor performance in AI model development and talent outflow phenomenon is interpreted to be influenced by a reduction in South Korea's investment scale. Last year, South Korea's private AI investment amounted to $1.39 billion (approximately 1.9 trillion KRW), ranking 9th among the surveyed countries. Considering that it was $3.1 billion in 2022, ranking 6th, the investment amount was halved in one year.
The data also confirmed an overwhelming gap among the leading groups. The United States, which ranked first with a private AI investment scale of $67.2 billion last year, significantly outpaced second-place China ($7.7 billion). China, which started with an investment of $620 million in 2013 and peaked at $23 billion in 2021, saw its investment sharply shrink to about one-third of its record over the following two years. The market evaluates that this was influenced by the U.S. semiconductor export controls against China.
Meanwhile, in terms of AI model development performance by company, U.S. IT companies including Google dominated the top ranks. Last year, Google released the most foundation models with 18 products, including Gemini. Meta (11), Microsoft (9), and OpenAI (7) followed. In the MMMU benchmark testing the capabilities of AGI (Artificial General Intelligence), Google's Gemini Ultra and OpenAI's GPT-4 led with accuracy rates of 59.40% and 56.80% respectively. However, both models fell short of the average human expert score (82.60%).
© The Asia Business Daily(www.asiae.co.kr). All rights reserved.
