As developers and publishers continue to pitch generative AI dialogue and neo-NPCs as the future of open-world video games, a growing counter-perspective from industry veterans highlights a fundamental disconnect over what makes game writing work.
The AI Pitch Versus Practice
Companies like Ubisoft, Saber Interactive, and start-ups such as Convai have argued that generative AI can solve the ballooning costs and ambitions of big-budget open-world video games. The premise is that infinite dialogue chains and AI-powered background characters can populate expansive worlds with minimal manual scripting.
In practice, however, early implementations in life sims like Seed or Inzoi have largely fallen flat. Rather than deep storytelling, the technology often results in awkward interactions or novelty experiments.
Why Veterans Call It a Conceptual Failure
Industry veterans have openly dismissed the technology. Larian's head writer described AI output as worse than their worst drafts, while Dragon Age creator David Gaider expressed exhaustion at the prospect of AI-generated stories. Cyberpunk 2077's quest director and id Software's Chris Hays have also voiced skepticism amid broader executive missteps regarding creative talent and automated substitutes.
Critics argue that procedural generation and AI barks miss the point of what makes background details memorable. Hand-crafted elements—such as town criers, combat barks, journal entries, and corporate advertisements in Cyberpunk 2077—create a deliberate sense of place and pacing. By contrast, procedurally generated worlds, such as those in Starfield, risk feeling perfunctory when stripped of hand-crafted text.