According to Kotaku, Nexon CEO Junghun Lee told Japanese outlet Game*Spark that “every game company is now using AI” and treating it as a survival strategy, while fellow South Korean publisher Krafton revealed an “AI first” plan to invest roughly $70 million in GPU clusters and infrastructure. The comments come amid controversy around Nexon’s own Arc Raiders, where design director Virgil Watkins claimed the game “in no way uses generative AI whatsoever” despite using text-to-speech models for all NPC barks, creating confusion about what actually counts as AI in game development. Embark Studios CCO Stefan Strandberg defended the approach by saying text-to-speech allows them to handle tedious repetition where voice actors might not find value, while Eurogamer’s review called the AI voices “aggressively mediocre and a stain on the experience.”
The corporate AI survival argument
Here’s the thing about Lee’s comments – they’re not really about technology. They’re about business strategy in an increasingly competitive market. When he says AI is a matter of survival, what he’s really talking about is efficiency and scale. Game development has become brutally expensive, and publishers are looking for any edge they can get. Text-to-speech for NPC barks? That’s basically replacing what would otherwise be expensive voice acting sessions for content that players might barely notice. But is that actually survival, or just cost-cutting dressed up as innovation?
Krafton’s massive $70 million AI investment shows this isn’t just about voice generation. They’re building infrastructure for what they call “agentic AI” across their entire company. That could mean everything from NPC behavior to procedural content generation to player support. The scale of that investment suggests they see this as fundamental to their future operations, not just a nice-to-have feature.
The “what counts as AI” problem
Now we get to the really messy part – nobody can agree on what actually constitutes “AI” in games. Arc Raiders’ design director says they don’t use generative AI, while simultaneously admitting they use machine learning during development and text-to-speech for voices. So which is it? The studio seems to be drawing a line between “generative AI” (which they claim not to use) and other forms of AI like text-to-speech and machine learning.
But players and critics aren’t buying these technical distinctions. When you hear obviously synthetic voices in a game, it feels like AI regardless of what specific technology is being used. Strandberg’s argument about using text-to-speech for “tedious repetition” where voice actors “may not see it as valuable work” raises another question – shouldn’t game developers be making sure ALL content is valuable, rather than outsourcing the boring parts to algorithms?
Not everyone’s on board
Meanwhile, the indie development community is pushing back hard against Lee’s assumption that everyone is using AI. Necrosoft Games, developers of Demonschool, publicly stated they “would rather cut off our own arms than do so” and emphasized their game is “100% human made.” That’s a pretty dramatic rejection of the AI-everywhere narrative.
And they’re not alone. The backlash against AI in creative fields has been growing, with many artists, writers, and developers concerned about being replaced by systems trained on their own work. The broader conversation around AI in games isn’t just about efficiency – it’s about artistic integrity, labor rights, and what kind of experiences players actually want.
Does AI actually improve quality?
Lee claims AI will increase the “average” quality across games, but Arc Raiders’ reception suggests otherwise. Eurogamer’s review specifically called out the AI voices as damaging the experience. So we have to ask – is this really about quality, or just about doing more with less?
The reality is that AI tools can be incredibly useful for certain tasks, but they’re not magic bullets. Poor implementation can actively harm player experience, as we’re seeing with Arc Raiders. And when companies try to hide their AI usage behind technical definitions, it just creates more distrust. Maybe the real survival strategy isn’t racing to adopt every AI tool available, but knowing when human creativity matters most.
