[Insight & Opinion] Legislative Discussions Needed on AI News Utilization and Revenue Sharing

Published 11 Sep.2023 12:37(KST)

This article compares the performance of generative artificial intelligence (AI) models Naver HyperCLOVA X and ChatGPT 3.5 in terms of high-level intellectual activities. The comparison involved assessing whether they provide academically useful answers to questions about social networking services (SNS) asked in Korean. CLOVA X produced answers consisting of two to three sentences citing a single domestic paper. Although the knowledge was recognized in academia, it was simple information not much different from portal search results. It did not access English literature at all. In contrast, ChatGPT cited six multinational papers and answered the SNS issue by dividing it into four categories in a lengthy response. Its depth and systematic approach stood out. Some of the proposed concepts were original. However, one of the references was fabricated.

"Tell me the fastest way to get from Geumcheon-gu to Mokpo." According to some tests, CLOVA X is useful for such questions specialized in everyday Korean life. However, in a comprehensive evaluation, CLOVA X falls short of ChatGPT.

The Korea Newspaper Association issued a statement regarding CLOVA X, saying, "Using news for AI training without proper grounds infringes on the copyrights and other rights that news organizations hold regarding their news." This reflects the media's concern that "after taking a significant portion of the news advertising market through portal search models, now AI models are trying to take the intellectual property of news organizations."

Thus, domestic generative AI development fails to demonstrate the ‘expertise’ competitive on a global scale or the ‘goodwill’ to grow alongside various sectors of society. Rather than leaving it to a specific company, the government and the National Assembly should enhance expertise in the generative AI field and establish policies and laws that implement goodwill.

Without AI, it is difficult to distinguish between the third industrial revolution of digitalization and the fourth industrial revolution of intelligence. In other words, AI will drastically change the world during the inevitable fourth industrial revolution. AI will be embedded in objects like glasses. Generative AI, which converses with humans through large language models (LLMs) that learn massive amounts of data and parameters to replicate human language processing, stands at the pinnacle. Sam Altman, the creator of ChatGPT, said, "ChatGPT is comparable to Gutenberg’s printing press."

The controversy that ChatGPT 4.0 performs worse than ChatGPT 3.5 is said to be temporary. Ultimately, visionaries aim to create artificial general intelligence (AGI) based on LLMs that embodies self-awareness and a body to realize general human intellectual activities. Some experts predict the AI tipping point?the moment of explosive growth?around 2030. According to investment bank Morgan Stanley, Microsoft, which owns ChatGPT, will apply this AI to office programs used by the majority. Such office AI will enable one person to achieve the productivity of an entire team. Companies and schools will have no choice but to purchase it. In this way, someone will dominate the world with AI. LLMs can also change the fate of the Korean language in the future. This is why national interest in securing competitiveness in generative AI is necessary.

The activation of generative AI will weaken portal search. The portal-centered media and knowledge industry ecosystem will face a new phase. There is a need for policy and legislative discussions on what AI’s revenue model is and how to share profits with news organizations and others when AI uses knowledge products like news for machine learning and provides answers. Korea is far behind the European Union (EU) in AI legislation. The nation must pay attention to the expertise and goodwill of generative AI.