Ex-NCSoft President: AI Is Essential to Game Development

With a doctorate in computational neuroscience from MIT and a 15-year tenure leading Korean developer NCSoft, Songyee Yoon’s perspective on artificial intelligence in gaming is deeply informed. Now heading the venture firm Principal Venture Partners, she invests in AI-native startups while reflecting on how the technology became indispensable at a major studio. Her journey underscores a fundamental shift: AI is now essential to game development.

Early in her NCSoft career, Yoon saw clear applications for AI that others dismissed. Her background in data science led her to propose using player data,login patterns, social interactions, and play logs,to predict player churn. The goal was to enable proactive interventions that could boost player retention, a critical metric for any live service game. While initially met with skepticism, this approach is now standard practice across the industry, demonstrating how data-driven insights can create tangible business value.

At NCSoft, AI’s integration became remarkably widespread. The company encouraged even non-technical staff, from daycare teachers to HR professionals, to learn basic programming like Python to query data platforms. An HR bot was deployed years before ChatGPT’s rise. Yoon notes that AI now touches nearly every facet of development: marketing trend prediction, monetization strategies, virtual goods packaging, NPC dialogue generation, adaptive enemy behavior, animation support, and even interactive music and lighting systems. It’s harder to find areas where AI isn’t used.

Yoon left NCSoft in 2023 to found PVP, driven by a conviction that a new wave of companies is emerging. She distinguishes between older firms undergoing digital transformation and truly AI-native companies built from the ground up with modern AI infrastructure. Just as digital-native companies like Google and Amazon defined the broadband era, she believes the next generation of leaders will be those fully leveraging the contemporary AI tech stack to create novel, significant value.

The current surge of investment and hype around AI does create concern about a potential AI bubble, Yoon acknowledges. A platform shift of this magnitude generates immense demand to fund the next winners, but the number of companies that will become enduring, generational successes is inherently limited. This mismatch between investment demand and viable supply can inflate valuations, especially when marketing hype overshadows the hard work of building substantive, lasting products.

To navigate this noise, PVP focuses on timeless investment theses. One key area is AI infrastructure,companies providing the essential tools and compute resources builders need. Another focuses on verticals rich in unstructured data, such as insurance, legal services, accounting, and healthcare. In these fields, AI can unearth insights invisible to the human eye, creating a compounding data advantage that acts as a durable competitive moat over time.

A common criticism is that AI can become a solution in search of a problem, leading to products users ignore. Yoon stresses that foundational business principles still apply: any product, AI-powered or not, must deliver clear utility and value. She recalls building an early machine learning credit-scoring model for a bank, a project that required physically transporting data tapes. The model succeeded because it solved a clear problem,assessing risk more accurately,not because it used AI. The technology is a means, not an end.

The use of AI for generative tasks like creating images and voices raises complex questions about artistic integrity and IP rights. Yoon predicts the internet will soon contain more AI-generated than human-generated content. The critical human role, she argues, will be in curation and orchestration,using AI tools to produce work that genuinely resonates with an audience. This requires a new form of artistry. She also emphasizes that human creativity cannot be fully replaced and calls for stronger legal frameworks to protect artists and their intellectual property, ensuring the data that trains future AI continues to be created.

The games industry has often been a laboratory for innovation, pioneering business models like virtual goods and community gamification. Yoon expects this trend to continue with AI, with gaming’s experiments foreshadowing broader societal adoption. The varied reception to different AI uses,such as the backlash against AI voices in Arc Raiders versus acceptance of its machine-learned enemy behavior,highlights a nuanced public conversation. Yoon sees AI fundamentally as a tool for augmenting human creativity, a means to express ideas that were previously limited by technical skill or coordination. Transparency about its use, she believes, allows users to make informed choices.

On the critical issue of ethics and training data, Yoon advocates for transparency. She compares it to food labeling, which protects consumers without stifling innovation. Understanding a model’s ingredients and limitations helps users navigate its outputs, acknowledging that AI systems carry inherent biases. A transparent approach, she argues, fosters consumer protection and a more sustainable, trustworthy future for the technology, ensuring competition thrives while users are safeguarded.