Global Games' Secret: AI
Why Localization Quality Defines Global
Players can assign a roommate NPC to chat with them and characters in up to 13 languages.
“What does this mean…?”
In the middle of a battle, a European player types something into the chat.
Moments later, an English message appears on the Korean player’s screen:
“There’s an enemy coming from the right!”
Their movements immediately sync. With mutual understanding, the exchange becomes strategy, and the strategy becomes seamless collaboration.
What began as awkward gameplay now flows like a well-rehearsed team.
This isn’t just a cinematic moment, it’s the reality made possible by AI-powered real-time translation.
In today’s MOBA and FPS games, players can fight, laugh, and play together—even if they speak different languages.
ℹ️ Language Meets Emotion: Blizzard’s Localization Philosophy
At Blizzard, localization is not just about translation—it’s a core part of their “Think Globally” philosophy.
Rather than just converting text, Blizzard adapts its game content to align with regional culture and sentiment, from storylines and humor to character tone and in-game slang. This emotional localization deepens immersion, making the experience feel authentic to each market.
“THINK GLOBALLY” – One of the Eight Core Values Engraved Under the Orc Statue at BlizzardThe Korean versions of <World of Warcraft> and <Overwatch> weren’t mere language swaps.
Curses and memes were rephrased with culturally relevant Korean expressions, while NPC speech patterns and quest titles were tailored to resonate with Korean sensibilities.
For example, when an NPC says, “Don’t piss me off” in the English version, the Korean version of <World of Warcraft> doesn’t translate it literally like “Don’t make me angry.” Instead, it becomes: “아, 진짜 성질나게 하네” (“Ah, seriously, you’re getting on my nerves”)
—a more natural fit for Korean emotional nuance.
This type of localization doesn’t just improve comprehension, it enhances emotional engagement, leading to longer playtime, higher ARPU, and stronger community loyalty.
Riot Games, which entered the Korean market after Blizzard, adopted a similar philosophy. League of Legends has maintained the No.1 PC café share in Korea for years by continuously investing in high-quality localization that preserves emotional authenticity.
ℹ️ Immersion Drives Monetization: The Business Case for Quality Localization
The business impact of high-quality localization is unmistakable.
According to Steam user review data, over 80% of players who rated localization quality positively also recommended the game. These titles saw retention and monetization rates up to 2–3 times higher than average.
<Genshin Impact> is a standout example.
With support of 13 languages in text and four in voice, the game delivered a seamless native-feeling experience across global markets. Korean players felt like it was made specifically for them, from the UI to voice acting. That emotional resonance translated into action, generating over $320M in revenue from Korea alone and $9B globally from 180M downloads.
<Witcher 3>, localized into 15 languages, also maintained long-term success, selling over 40 million copies worldwide.
In contrast, <Assassin’s Creed III> became a cautionary tale. A pivotal line — “Damn you, fire!” — was mistranslated as “It’s on fire!” in Korean. This small error sparked user confusion, churn, negative word-of-mouth, and a drop in ratings, illustrating how localization missteps can erode brand trust.
Localization is not just a user-facing feature. It defines how players emotionally connect with a game — and that connection directly influences business outcomes.
ℹ️ What AI Translation Has Changed: Immersion, Emotional Bond, and Play
Localization used to be a pre-launch task — translating text before going live.
Now, AI enables real-time translation of chat, voice, and commands during gameplay.
Players from different language backgrounds can now share tactics and coordinate in real-time. Cultural barriers are fading. This not only increases playtime but also increases purchasing like season passes and skin items.
<Fortnite> reflects this shift, actively experimenting with AI-driven conversational content. However, as of July 2025, its AI chat capabilities are not universally applied across all player interactions.
For instance, the Darth Vader AI chat introduced during the Star Wars season was limited to interactions with specific NPCs, not general player-to-player conversations. A separate creator-focused feature using Unreal Editor is also in development, enabling AI-powered NPC dialogue, but again, not yet extended to real-time player chats.
In short, <Fortnite> use of generative AI remains confined to specific seasons, content types, and creator tools. Still, even these limited applications have proven effective in enhancing immersion and emotional engagement.
ℹ️ AI That Talks Back: <Play Together> by HAEGIN
A notable example from Korea is HAEGIN’s <Play Together>, which expands AI use from simple translation to real-time conversation.
Play Together is a global social mobile game (excluding China) that supports 13 languages. After a player remarked, “I wish the NPC in my house could talk to me like a real person,” the development team responded by introducing an AI-powered roommate chatbot.
Players can assign a roommate NPC to chat with them and characters in up to 13 languagesInitially, HAEGIN tested GPT-3.5 and GPT-4-based models but struggled with high costs and low retention. After migrating to Claude 3 Haiku, usage increased 3.6 times. While overall cost rose, token unit costs were cut in half. Most notably, paid chat point usage surged 16-fold, significantly boosting ROI.
Yet beyond the metrics, the emotional response stood out: players described these NPC interactions as “real social relationships.” They felt, “The game remembers me.” That emotional bond converted directly into monetization.
ℹ️ How Real-Time Translation Works — and Where to Apply It
Real-time translation isn’t just about converting words into another language. It replicates human-like understanding of context and intent. Here’s how the process works:
1. Speech-to-Text (STT): The AI listens to the user’s voice and transcribes it into text.
Example: When a player says, “Watch your back!”, the system converts it to text.
2. Context Analysis: Instead of analyzing each word individually, the AI understands the context in which the words were spoken within a gaming situation.
3. Natural Translation Generation: Rather than doing a literal translation, the translation is generated in a manner that reflects the actual language used by players.
Example: "Rush mid, I'll frag" → "Let's rush mid, I'll take care of them" (Korean translation).
4. Text/Voice Output: The translated result is displayed in the chat or played as speech, allowing players to instantly understand teammates across languages.
This entire process involves a combination of AI models, speech recognition, and real-time processing systems. Recent advancements have enabled some tasks to run on local devices instead of remote servers, improving speed and responsiveness.
However, applying this to all conversations is challenging due to cost and resources. AI translation requires routing messages to the cloud for processing — so with more users, processing demand and costs rise exponentially. That’s why most games adopt the following operational strategies:
Selective Use in Key Moments: Real-time translation is prioritized for global co-op, esports match, or VIP user interactions
Caching Common Phrases: Frequently used messages like “Help me!” or “Come here!” are pre-translated and cached for faster delivery
Hybrid Translation Architecture: Basic phrases use open-source translation tools, while more nuanced messages leverage AI to balance cost and quality
In summary, the success of real-time AI translation depends less on the technology itself and more on the operational strategy of 'where and how' to utilize that technology. Game-ready AI solutions must be built with operational efficiency and user experience in mind.
♻️ Final Thoughts
AI translation is no longer just a technical feature.
It’s a tool for experience design — one that builds emotional connection, sustains immersion, and strengthens trust between players and the game.
Language is more than text or voice; it’s an emotional message from the brand.
When players feel understood, they stay longer, engage deeper, and convert more.
Now that localization is covered, the next post will explore how AI is transforming marketing campaigns, lightening the workload while driving tangible outcomes.
※ Disclaimer: This content reflects the author’s personal views and includes only publicly available examples. It does not represent the official position of any company mentioned
