In the complex, high-stakes world of modern game development, efficiency is survival. Rebel Wolves CEO and veteran designer Konrad Tomaszkiewicz has ignited a conversation after revealing his studio’s pragmatic use of AI-generated voice acting as advanced placeholders during the early development of their upcoming open-world vampire RPG, The Blood of Dawnwalker. The strategy, born from his experiences on The Witcher 3: Wild Hunt and Cyberpunk 2077 at CD Projekt, aims to solve a perennial and costly problem in narrative-heavy games: finding major design flaws only after expensive, multi-language voice recording sessions are complete. While this approach is presented as a time-saving measure, it lands squarely in the industry’s heated debate about the ethical contours of using artificial intelligence in creative processes.<\/p>\n
To understand Rebel Wolves’ decision, one must grasp the monumental scale of a project like The Blood of Dawnwalker. It is described as having an amount of storyline “similar or even more than in The Witcher 3,” a game legendary for its dense narrative and sprawling side quests. Tomaszkiewicz details that final voice recordings occur in at least eight languages: English, Polish, Spanish, French, German, and Japanese among them. The financial and logistical weight of these sessions, involving scores of professional actors, directors, and sound engineers, is immense. Once these recordings are locked in, any subsequent change to a quest’s flow, dialogue, or context becomes prohibitively expensive, often requiring re-booking actors and re-recording entire sections.<\/p>\n
The core issue, as Tomaszkiewicz explains, is that designers and writers often fail to spot narrative or pacing problems when simply reading text on a page or in a script. “When you’re reading stuff, you sometimes don’t recognise the problems you have in the game,” he stated. It’s only when dialogue is heard in context, with tone, timing, and placement within the game world, that issues like awkward phrasing, lore inconsistencies, or confusing quest instructions become apparent. Traditionally, studios might use placeholder text, basic text-to-speech, or even have developers record “scratch” vocals, but these methods often lack the nuance needed for proper evaluation in a complex RPG.<\/p>\n
Rebel Wolves’ proposed solution was to use generative AI to create convincing, albeit temporary, voice performances for the entire game during its early development phase. This allowed the team to integrate full dialogue into game builds much sooner. Playtesters and developers could experience quests with voiced characters, identifying logical gaps, pacing issues, and narrative weaknesses long before a single human actor stepped into a studio booth.<\/p>\n
“We used AI to generate the first voices for the game, to quite early in the process find out all the problems we have, before we recorded it with the actors,” Tomaszkiewicz said. He emphasized this was purely for internal iteration. The AI audio was a prototype, a sophisticated stand-in to be discarded. By finding and fixing these problems iteratively using AI voices, the studio aimed to enter the human recording phase with scripts and quest designs that were far more polished and locked down. The goal was a net saving: invest in AI tools upfront to avoid massive, reactive costs later. Tomaszkiewicz stresses that every voice line in the final game is performed by a human actor, with all AI-generated placeholder audio completely erased.<\/p>\n
While the logic for efficiency is clear, Tomaszkiewicz’s revelation leaves several critical questions unanswered, which form the heart of the ethical debate. The CEO did not specify which AI technology was used, a detail with significant implications. Was it a proprietary, ethically-sourced tool trained only on consented, compensated data? Or was it a more common large language model (LLM) or voice synthesis platform trained on vast, indiscriminate datasets scraped from the internet, potentially including the uncompensated work of voice artists?<\/p>\n
The voice acting community and its unions, such as SAG-AFTRA, have spent years campaigning for protections against the unauthorized use of their performances to train AI systems or create “digital replicas.” Their landmark 2023 strike against major game publishers resulted in new agreements specifically addressing AI, demanding transparency, consent, and compensation. From this perspective, even the use of AI for placeholders is problematic if the underlying technology was built on what they view as intellectual property theft\u2014the unauthorized use of human voices to create a synthetic substitute.<\/p>\n
Beyond the immediate data source issue lies a broader industry fear: normalization. Critics argue that legitimizing AI for “harmless” placeholder work creates a precedent and infrastructure for its use in darker applications. The concern is that cost-conscious executives, having seen AI used successfully in pre-production, might later ask: “If AI voices were good enough for testing, why can’t they be good enough for the final game, especially for minor characters? It would save even more money.” This is not hypothetical. Instances have emerged of indie developers attempting to replace hired human voice actors with AI-generated performances post-contract, a move widely condemned by the creative community.<\/p>\n
This slippery slope challenges the very defense Tomaszkiewicz offers. He positions AI as a tool to make human artistry more efficient and protected<\/strong> from costly re-dos. However, skeptics see it as the first step in a process that could ultimately marginalize or replace<\/strong> that same human artistry. The line between a placeholder and a replacement becomes dangerously thin when the technology to create a passable synthetic performance is readily available and normalized within development pipelines.<\/p>\n The debate naturally leads to an examination of alternative methods. How have other studios handling massive RPGs solved this iteration problem without stepping into the AI minefield? Some continue to rely on developer-recorded scratch tracks, accepting their limitations. Others invest in early, smaller-scale recording sessions with key actors to prototype critical narrative moments. There is also a school of thought that advocates for stronger, more disciplined pre-production writing and design reviews to catch issues before any form of voiceover, even placeholder, is needed.<\/p>\n However, as Tomaszkiewicz notes from his CD Projekt experience, the sheer scale of something like The Witcher 3 or Cyberpunk 2077 can overwhelm these traditional methods. The pressure to iterate quickly and the catastrophic cost of post-recording changes create a powerful incentive to seek technological solutions. The question becomes whether the industry can develop or endorse ethical AI tools designed specifically for this purpose\u2014tools built with transparent, compensated data sets and clear contractual firewalls that prevent their output from ever being used in a final product.<\/p>\n For the practice described by Rebel Wolves to gain wider acceptance, it will require a level of transparency not yet provided. Studios choosing this path may need to publicly disclose the AI tools they use and the data sourcing policies of those tools, much like they would highlight their game engines or middleware. Partnerships with voice actor unions to develop agreed-upon, ethical placeholder systems could offer a middle ground, ensuring the technology supports rather than undermines the profession.<\/p>\n Furthermore, the definition of “placeholder” must be contractually and technically ironclad. The AI-generated audio files should be watermarked, stored separately, and automatically deleted from build systems after a certain phase, ensuring they cannot be accidentally or intentionally used beyond their intended scope. Developing these guardrails is as important as the efficiency argument itself.<\/p>\n Konrad Tomaszkiewicz’s defense of AI voice placeholders in The Blood of Dawnwalker highlights a genuine, complex tension at the intersection of art, technology, and commerce. It is a testament to the daunting challenge of producing vast, story-driven games where voice acting is not a garnish but a central pillar. While the pursuit of efficiency is undeniable, the gaming industry’s journey with AI is proving that no tool is morally neutral. The legacy of this decision, and others like it, will depend not just on the time and money saved, but on the industry’s collective commitment to protecting the human creativity that remains the soul of the medium. The success of The Blood of Dawnwalker will ultimately be judged by its final, human performances, but the conversation around how it reached that point is a critical chapter in the ongoing story of game development itself.<\/p>\n","protected":false},"excerpt":{"rendered":" In the complex, high-stakes world of modern game development, efficiency is survival. Rebel Wolves CEO and veteran designer Konrad Tomaszkiewicz has ignited a conversation after revealing his studio’s pragmatic use of AI-generated voice acting as advanced placeholders during the early development of their upcoming open-world vampire RPG, The Blood of Dawnwalker. The strategy, born from […]<\/p>\n","protected":false},"author":5,"featured_media":86493,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"fifu_image_url":"https:\/\/cards.overcentral.com\/cards\/en\/38616.png","fifu_image_alt":"Blood of Dawnwalker CEO defends AI voice placeholder use to save time","footnotes":""},"categories":[349],"tags":[],"class_list":["post-38594","post","type-post","status-publish","format-standard","has-post-thumbnail","category-articles"],"fifu_image_url":"https:\/\/cards.overcentral.com\/cards\/en\/38616.png","fifu_image_alt":"Blood of Dawnwalker CEO defends AI voice placeholder use to save time","_links":{"self":[{"href":"https:\/\/overcentral.com\/en\/wp-json\/wp\/v2\/posts\/38594","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/overcentral.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/overcentral.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/overcentral.com\/en\/wp-json\/wp\/v2\/users\/5"}],"replies":[{"embeddable":true,"href":"https:\/\/overcentral.com\/en\/wp-json\/wp\/v2\/comments?post=38594"}],"version-history":[{"count":0,"href":"https:\/\/overcentral.com\/en\/wp-json\/wp\/v2\/posts\/38594\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/overcentral.com\/en\/wp-json\/wp\/v2\/media\/86493"}],"wp:attachment":[{"href":"https:\/\/overcentral.com\/en\/wp-json\/wp\/v2\/media?parent=38594"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/overcentral.com\/en\/wp-json\/wp\/v2\/categories?post=38594"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/overcentral.com\/en\/wp-json\/wp\/v2\/tags?post=38594"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}Alternative Methods and The Human Element<\/h2>\n
The Path Forward: Transparency and Ethical Guardrails<\/h3>\n