The debate surrounding Artificial Intelligence’s integration into video game narratives is no longer theoretical. As tools like generative AI become increasingly powerful and accessible, the games industry faces a critical crossroads where the promise of unprecedented player agency and dynamic storytelling collides with profound ethical concerns. At the heart of this debate are fundamental questions about creative integrity, the value of human authorship, and the future role of writers and designers. To explore these complex issues, games writer Daniel Griliopoulos (Total War: Warhammer 3) and Thomas Keane, co-founder of the AI-focused studio Meaning Machine (Dead Meat), engage in a deep discussion on whether AI can find an ethical, creatively valid role in shaping the stories we play.
The Current State: Ethical Problems and Creative Imperatives
Both debaters begin from a shared understanding of AI’s significant ethical challenges. Daniel Griliopoulos frames them as the “three Ps”: Power, Plagiarism, and job losses. He notes that while pathways past these issues might exist, most current implementations do not take them. The rapid, often uncritical adoption by early enthusiasts, he suggests, frequently sidelines moral considerations. However, the reality is undeniable: AI use is already widespread. Thomas Keane fully acknowledges these problems but argues for a proactive, engaged response from the creative community. “We can’t just back away from it,” he states, emphasizing the need for developers to define ethical standards themselves and “carve that new reality,” rather than abdicating responsibility to those outside the craft.
The Human Hand vs. the AI Slop: Defining a Role for AI
When asked about AI’s specific role in narrative, Keane is unequivocal: left alone, AI generates “soulless incoherent slop.” The core of Meaning Machine’s philosophy, therefore, is not to replace humans but to enable them. “The only thing that can drive and create quality is the human hand,” he asserts. Their approach rejects using AI for production shortcuts or cost-cutting. Instead, they employ it as a runtime technology to power new mechanics, creating experiences characterized by “unprecedented flexibility and adaptability” alongside player freedom. Griliopoulos offers a broader philosophical counterpoint, questioning whether human vision is inherently superior on grounds of plagiarism, given that all human art is built upon learned language and prior works. He sees potential in AI for new possibilities but remains skeptical that they currently surpass the quality and curated intentionality of hand-designed content.
The Hybrid Model: Infusing Emergence with Handcrafted Content
The discussion then focuses on the practical method of achieving quality. Keane describes his studio’s process as “bullying AI into doing something interesting.” This involves rigorously constraining generative systems with a foundation of hyper-specific, hand-authored content created by their writers. For instance, an AI’s generic response about weather is transformed by linking it to a deeply personal, human-written memory. This “collision between something very flexible, and something very specific that is human authored” is where Keane believes compelling, unique experiences emerge. This model requires extensive writing, not less. “We actually see the birth of AI and video games as not the death of the writer, we see it as the era of the writer,” Keane argues, positioning high-quality written content as the decisive competitive advantage.
Griliopoulos raises concerns about current AI’s reliability and tendency to “break the frame,” but acknowledges that systems like Keane’s aim to address this. His personal stance, however, is rooted in the craft itself. He contrasts the “corporate side” interested in the end product with the artist’s motivation: “I don’t want to get Midjourney to make me a piece of art, I want to paint the painting.” For him, the process of writing—even the “boring bits”—is intrinsically valuable, a perspective that suggests a tension between authorship and curation in an AI-assisted future.
Constraints, Quality, and Player Response
A key technical and creative hurdle is constraint. Griliopoulos points out that making AI characters stay within their narrative lane without producing nonsensical or immersion-breaking output is still a significant challenge. Keane agrees that raw AI quality is inadequate but believes their hybrid system can achieve parity with handcrafted content while adding flexibility. He cites independent, longitudinal research from the University of Bristol, which has studied their demos for three years, finding positive player responses in terms of creative freedom, narrative immersion, and personal gratification. This, he stresses, does not invalidate traditional linear storytelling but proves there is distinct value in this new format, akin to the difference between a scripted game and a tabletop role-playing session.
The Future Infiltration: Mainstream Adoption and Industry Silos
Looking ahead, the debaters speculate on how AI will permeate the industry. Griliopoulos foresees initial adoption in spaces with “less ethical constraints,” such as VR, mobile, and some MMOs, drawing a parallel to the gradual mainstream acceptance of monetization models like loot boxes. He believes the “core gaming space” that values hand-curated, premium experiences will resist longer. Keane views AI as another catalyst for the industry‘s history of “aggressive reinvention,” leading to new types of games rather than eradicating old ones. He reveals that many “reputable names” are currently experimenting behind closed doors, awaiting a perceived quality bar before going public. This period of clandestine development, he suggests, will eventually bubble up into the mainstream.
The Ethical Pathway: Legislation, Compensation, and On-Device Power
Returning to the foundational ethical concerns, both thinkers outline potential solutions. Griliopoulos sees a route through copyright reform, possible legislation, or retraining AIs on non-copyrighted material. For voice and artistic style, he suggests models where individuals are paid for the use of their likeness or style, akin to a buyout. The critical issue of energy consumption, he notes, is “only creeping up.” Keane strongly advocates for on-device processing as the answer to the power problem, halting the environmentally destructive “arms race” of ever-larger model training. He agrees that legislation must be rethought and creators fairly compensated.
Ultimately, they converge on the most immediate ethical action: engagement. Keane’s mission is to “carve out a role for humans in a future where AI is a serious force,” creating a compelling case against the boardroom desire to simply eradicate workforces. Griliopoulos concurs, framing engagement as an ethical imperative: “if you don’t, the worst people in the world will.” Their debate concludes not with a simple verdict but with a call for the creative community to steer the technology toward a future that values both innovation and the irreplaceable human element at the heart of storytelling.
