David Gaider, the former lead writer on the first three Dragon Age games, warned that generative AI tools could make game development "frustrating as hell" for future teams. His comments, published in a GamesRadar feature, arrive as studios increasingly test AI in creative workflows and executives push to integrate the technology despite its well-documented flaws.
AI's inconsistency creates a troubleshooting nightmare
Gaider pointed to a core problem that writers and narrative designers will recognize immediately: AI output lacks consistency. When a tool produces something that doesn't work, the developer has to go back and fix it without any clear understanding of why the model generated that result in the first place. "It's not ready for prime time," Gaider said. "There's just a lot of executives who really, really want it to be."
That black-box nature means appraising, troubleshooting, and cleaning up AI-generated work becomes a grind. Instead of saving time, teams end up wrestling with unpredictable output. For writers who already spend significant energy on revision and editorial polish, the prospect of debugging a model's decisions on top of that work is a hard sell.
The junior developer pipeline problem
One of the most common defenses of AI in game development is that it can handle rote, entry-level tasks - the kind of work often assigned to junior staff. Gaider pushed back on that framing directly. "How are we going to train up the next generation of devs if we eliminate every entry-level task?" he asked.
Those small, unglamorous assignments are how new writers and designers learn the craft. Removing them doesn't just cut costs in the short term; it severs the pipeline that produces senior talent. The industry already struggles with on-ramps for early-career creatives, and automating that work away would narrow the path further.
Ethical concerns and player skepticism
Gaider also addressed the use of AI for early prototypes and placeholder art, noting that artists haven't agreed to have "their data pillaged." That unease is shared by players. The reaction to the new Crazy Taxi game showed a vocal segment of the audience rejects AI's creative applications outright.
Other developers quoted in the same feature echoed similar concerns. David Szymanski, creator of Iron Lung and Dusk, said he is "not categorically against AI as a whole technology" but finds it a bridge too far to "hand wave all the ethical concerns about plagiarism, environmental impact, and job security." Marvel Rivals executive producer Danny Koo said his team avoided AI art tools specifically to ensure the game's assets weren't "poisoned" by plagiarism risks.
These aren't fringe worries from a handful of skeptics. They reflect a growing, industry-wide conversation about where the line sits between useful Generative AI and LLM tools and tools that undermine the people they're supposed to assist.
Why this matters for writers
The push to automate entry-level tasks hits writers harder than most because the writing profession already has few formal apprenticeship structures. Junior roles - drafting dialogue, writing item descriptions, building quest logs - are where writers develop voice and learn how narrative interacts with systems. If those roles shrink, the entire discipline loses its training ground.
Gaider's consistency complaint also resonates for anyone who has edited AI-generated text. The time spent diagnosing why a model made a strange choice often cancels out the time saved by generating it. For writers, the real question isn't whether AI can produce words. It's whether the output is reliable enough to trust without a second full pass - and right now, the answer from people who have actually tried to ship work with these tools is a firm no.
Writers looking to understand where AI fits into their workflow - and where it doesn't - can explore resources on AI for Writers that cover the practical and ethical dimensions of these tools without the executive hype.
Your membership also unlocks:
