The Cultural Arts Division of the Korea Game Society, Myongji College, and OGQ held a seminar titled 'The Age of AI: Concepts and Issues in Cultural Data Sharing Systems' on January 23rd at the Korea Creative Content Agency in Gwanghwamun, Seoul.

This seminar focused on the construction, sharing, and accountability structures of cultural data for AI training—a topic that has gained massive traction following the explosion of generative AI. Discussions spanned public data policy, European cultural data governance, and the thorny issues of AI ethics and bias.

Lee Kwon-soo, Team Leader of the Cultural Big Data Team at the Korea Culture Information Service, kicked things off with a presentation on 'Strategic Cultural Data Construction for Korea's Sovereign AI.' He didn't mince words: "While current AI can process Korean, it fails to truly grasp the context and narratives embedded in our culture." He argued that data needs to be 'explainable,' going beyond raw facts to include expert verification and context. Citing ongoing projects like the encyclopedia of folklore, he emphasized that we need high-resolution sources, precise metadata, and rigorous multilingual verification. "The public sector must lead by establishing the infrastructure and reliable standards that the private sector simply cannot manage alone," he noted.

Eun-jin Lee, a board member at the Game Rating and Administration Committee and a professor at Myongji College’s Department of AI Game Software, shifted the focus to Europe. In her talk, 'Key Issues in Cultural Data Sharing: Lessons from Europe,' she pointed out that the battleground for cultural dominance has moved from simple content production to the underlying logic of data training. "Culture is no longer just an end product; it has become training data. In the AI era, cultural sovereignty isn't about how far you've spread your content, but rather the standards and structures upon which your AI is built," she argued.

She specifically analyzed Europe's Gaia-X and data space initiatives as institutional responses to the chaos surrounding data origin, accountability, and rights. According to Lee, the takeaway isn't about copying European regulations, but understanding how they’ve engineered a system based on trust and clear order.

Instead of dumping all cultural data into a central bucket, Europe has focused on a decentralized model where stakeholders retain their data while agreeing on common rules and accountability frameworks. Lee noted that this approach treats data sharing as a governance issue rather than a technical one. "Korea must also shift its perspective," she stated, "moving away from simple open-data policies toward designing an order that protects our cultural and cognitive sovereignty."

The final speaker, Shin Chul-ho, an adjunct professor at Yonsei University and CEO of OGQ, tackled 'AI Ethics and Bias: Standards and Directions for Cultural Data Governance.' He argued that AI bias isn't a mere technical glitch but a structural failure in how data is curated. Analyzing biases in AI-generated images and text—ranging from race and gender to lifestyle—he remarked, "The so-called 'neutrality' of AI is often just a byproduct of Western-centric data and safety rules."

He warned that training AI on data with murky copyright status could undermine creator rights and erode trust in K-Content. He called for a national-level ethical dataset and robust filtering systems to catch bias and copyright infringements right from the training phase.

The general discussion reached a consensus: cultural data sharing is not just about opening up data sets; it’s a governance challenge that requires clear rules on access, usage, and accountability.

Participants agreed that the future competitiveness of K-Culture in the AI era depends on securing the right training structures and trust frameworks, necessitating a collaborative effort between the public sector, private companies, and academia.

If you missed the event, you can catch the full seminar on the Korea Game Society's official YouTube channel.