Embark Studios CEO: AI Won't Replace Workers on Arc Raiders
The head of Embark Studios has publicly addressed growing concerns about artificial intelligence in game development. Patrick Söderlund, CEO and founder of the studio, emphasized that their AI integration is not intended to reduce human involvement in creating titles like Arc Raiders.
In a conversation with GamesBeat, Söderlund made his position clear. "We don't use artificial intelligence to not have to hire people or replace people or job groups." The statement comes as the gaming industry grapples with how to balance efficiency gains with workforce stability.
According to the WN Hub report, Söderlund explained that AI serves as a tool to enhance efficiencies, particularly in situations requiring rapid updates to a game. This distinction matters—AI as augmentation versus AI as replacement represents fundamentally different business philosophies.
The studio maintains ongoing contractual relationships with voice actors. Söderlund stated, "We will pay for their voices, and sometimes using an artificial voice gets us to update the game a lot faster." The physical reality here is telling: actors still show up to recording booths, still get paid, still deliver performances. The AI layer sits on top as an efficiency tool, not a substitute.
Here's where things get interesting. Despite the assurances about not replacing workers, Arc Raiders launched in October with AI-generated dialogue that drew criticism from players. The game achieved record success for Embark's parent company, Nexon, but the text-to-speech implementation created friction in the player experience.
According to Engadget's coverage, Embark Studios subsequently replaced some AI-generated voice lines with professional actors after the successful launch. Söderlund admitted there is a "quality difference" between AI and human performance. "A real professional actor is better than AI; that's just how it is."
This reversal reveals something important about the current state of AI in games. The technology works for rapid iteration and placeholder content, but it doesn't yet match the emotional nuance of human performance in narrative-heavy moments. Players can tell the difference when a character's voice lacks the subtle inflection that comes from genuine acting.
The studio clarified that no generative AI was used in creating the game's visuals. This is a meaningful boundary—AI-assisted voice work is one thing, but replacing concept artists or 3D modelers crosses into different territory. The distinction between tool and replacement remains crucial for industry observers.
Söderlund credited AI, along with strategic investments in tools and exceptional staff, for enabling weekly game updates for their free-to-play shooter The Finals. "We couldn't have built the games that we've built or service the games that we've done," he explained. The human element remains central to game development, not fully automated processes.
It's worth noting the practical reality here. Arc Raiders peaked at nearly half a million users on Steam. The game's breakout success was still marred by its use of text-to-speech AI. Players noticed. They complained. The studio responded by re-recording lines with the original voice actors whose voices trained the AI system.
This creates an awkward middle ground. The studio pays actors for their time in the recording booth and will "continue to bring many of them back as we carry on updating the game." But some AI-generated lines remain in place. The decision isn't binary—human or machine—but rather a hybrid approach that prioritizes quality where it matters most.
From a business perspective, this makes sense. AI handles the monotonous tasks, allowing human resources to focus on meaningful creative work. The technology streamlines processes that would otherwise consume excessive time. But when immersion is on the line, human performance wins. (This is the moment where studios realize their cost-saving measures actually cost them player trust.)
The broader implication extends beyond Embark Studios. As AI tools become more sophisticated, the question shifts from whether to use them to where they fit best in the development pipeline. Voice acting for essential narrative moments? Human. Rapid iteration on non-critical dialogue? AI might work. The line keeps moving.
Whether this hybrid model becomes industry standard remains uncertain. The technology continues to improve, but so does player expectation. Studios that treat AI as a replacement rather than a tool risk the same backlash Embark encountered. The market has spoken—quality matters more than efficiency when it comes to core player experience.
For now, the real question isn't whether AI belongs in game development. It's whether players will continue to accept it in their games. Embark Studios found out the hard way that some things can't be automated without consequences. Whether users actually pay for the difference remains the real question.
Artūras Malašauskas is an AI Systems Integrator with 20+ years of production-grade web engineering experience. He has designed, shipped, and scaled enterprise Python/PHP systems for logistics, SaaS, and public-sector clients. For the past year, he has focused exclusively on AI integrations: deploying open-source LLMs, building generative media pipelines (image, audio, video), and engineering multi-agent workflows for real production environments. His standard: reproducibility, security, cost-efficient inference—no vaporware. He documents and evaluates emerging AI tooling, separating verified capabilities from marketing noise. Technical editor at: muza-ai.eu, ai-verslas.lt, ai-naujinos.lt Connect on LinkedIn
Artūras Malašauskas is an AI Systems Integrator with 20+ years of production-grade web engineering experience. He has designed, shipped, and scaled enterprise Python/PHP systems for logistics, SaaS, and public-sector clients. For the past year, he has focused exclusively on AI integrations: deploying open-source LLMs, building generative media pipelines (image, audio, video), and engineering multi-agent workflows for real production environments. His standard: reproducibility, security, cost-efficient inference—no vaporware. He documents and evaluates emerging AI tooling, separating verified capabilities from marketing noise. Technical editor at: muza-ai.eu, ai-verslas.lt, ai-naujinos.lt
Comments