Richard Garriott Says AI Has Taken Shroud of the Avatar's Bug Count to "Zero"

Richard Garriott has made a striking claim about the present state of Shroud of the Avatar: the long-running online RPG now reaches a bug count of "zero" every day. Speaking during a recent appearance on the TechJesse podcast, the Ultima creator credited artificial intelligence tools with dramatically accelerating the process of finding and fixing issues in the game.

The claim is especially notable because Shroud of the Avatar has been available for eight years and has carried a difficult reputation among parts of its player community. Like many persistent online games, it has continued to receive updates over time, and those updates can introduce fresh technical problems even when older ones have been resolved. Garriott's comments suggest that AI is now being used to keep pace with that ongoing maintenance burden rather than simply assist with the initial creation of new content.

Garriott framed the subject as controversial from the outset. He said AI is helping development move faster and requires fewer people and less time to make a game, a position that places him in the middle of one of the industry's most heated debates. Game development teams have increasingly adopted AI-driven tools for tasks ranging from testing and internal workflow support to asset generation and programming assistance. At the same time, those changes have intensified concerns about job displacement, training data, copyright, creative ownership, and the environmental costs associated with expanding AI infrastructure.

From a Persistent Backlog to Rapid Fixes

Garriott described a much different development reality for Shroud of the Avatar before the team's increased use of AI. He said the game historically maintained a backlog of roughly a thousand bugs at any given time. Some lower-priority issues, he said, remained unresolved for three to five years because the team had to direct its limited time toward more urgent problems.

For useful background on this topic, read Richard 'Levelord' Gray, Influential Duke Nukem 3D and Quake Level Designer, Dies at 68.

That sort of triage is familiar across the games industry. A bug list is not always a simple measure of whether a game is good or bad; teams must decide which issues break progression, damage performance, affect many players, compromise online systems, or can be safely left for later. Small visual defects, uncommon interactions, and edge-case problems can remain in a tracking system for a long time when a studio is balancing patches, new features, support work, and future content.

Garriott says AI has changed that equation for the game. In his description, newly discovered issues are now fixed within an hour of being found. He used Shroud of the Avatar as his main example when arguing that AI can handle a significant portion of the labor involved in maintaining a game's technical stability.

"Shroud of the Avatar, for example, now has zero bugs, every day. Zero."

It is a bold statement, and one that naturally depends on how a studio defines, detects, prioritizes, verifies, and closes bugs. Any actively updated online game can receive new reports as players explore systems in unexpected ways, as server conditions change, or as new patches alter existing interactions. Garriott himself connected ongoing updates with the possibility of new issues appearing. His broader point was that AI-assisted workflows can allow the team to clear those issues at a speed that was previously impractical.

AI's Role in Development Is Still Contested

Garriott did not present AI as an uncomplicated solution. He acknowledged concerns over whether AI systems may reproduce or draw improperly from existing work, as well as questions around voice technology and the people whose voices could be used in training. He said ethical AI use is important and stated that his team is trying to use the technology responsibly, while also conceding that these are not problems he has personally solved.

Those qualifications matter because "AI in games" covers a wide range of tools and practices. A system that sorts player reports, identifies repeatable steps behind a crash, or helps check game data for errors raises different questions than a system used to generate final art, voice performances, writing, or code. The impact on workers also differs depending on whether a tool helps existing staff reduce repetitive work or is used as a reason to reduce headcount and outsource fewer tasks.

Garriott's comments touch directly on that tension. His argument for AI is rooted in efficiency: fewer people and less time are needed to make and maintain games. For studios under financial pressure, that proposition is plainly attractive. Yet for developers, artists, writers, quality assurance staff, performers, and other workers, efficiency claims can also sound like a warning that their roles may be reduced, reshaped, or eliminated.

Quality assurance is particularly relevant to Garriott's example. Testing a game involves more than locating broken behavior. Testers document issues, establish reliable reproduction steps, judge severity, check fixes, consider side effects, and bring a human understanding of player expectations to the process. Automated systems can be valuable for repetitive validation and broad technical coverage, but the industry debate remains centered on where automation genuinely assists people and where it is used to replace expertise that still matters.

Humans for Art, Machines for Repetitive Work

Garriott's preferred division of labor is relatively clear. He said humans should create the art, while machines can help with the work of ensuring that the details are correct and the resulting game is stable, fast, and enjoyable to play. That places AI in a support role for technical consistency and production speed rather than presenting it as the primary author of a game's creative identity.

In practice, the boundary may not always be so neat. Technical decisions influence design, and creative work often depends on tools and pipelines that are deeply connected to engineering. A bug fix can change the feel of combat or movement; performance work can affect visual ambition; automated content tools can shape how designers build quests and encounters. Still, Garriott's formulation reflects a view held by many developers: automation may be most defensible when it removes tedious, repetitive tasks and leaves the expressive, judgment-heavy work to people.

For an online RPG with years of accumulated systems, that assistance could be substantial. Persistent games combine character progression, combat rules, item interactions, world events, user interfaces, multiplayer services, and frequent updates. Each addition may interact with older features in ways that are difficult to predict. The more quickly a team can identify and validate technical issues, the more time it can potentially spend on design, community support, and new content.

A Long Road for Shroud of the Avatar

Shroud of the Avatar occupies an unusual place in Garriott's career. His legacy as a foundational figure in computer role-playing games is firmly tied to the Ultima series, while his more recent projects have generated considerably more debate. The MMO has continued to be updated years after launch, but its development history and reception have remained subjects of discussion among RPG players.

That history makes the "zero bugs" assertion likely to draw close attention. It is not simply a technical boast; it is also a statement about whether a mature live game with a reputation for problems can be maintained differently through modern tools. Players will ultimately judge that claim through their own experiences with stability, performance, progression, and the arrival of future updates.

Garriott's remarks arrive as AI continues to become a central issue for the entire games business. Studios are looking for ways to shorten production cycles and manage escalating costs. Workers are pressing for safeguards around consent, credit, training data, and employment. Players, meanwhile, are asking whether the technology will lead to more polished games, less distinctive work, fewer jobs, or some combination of all three.

For Garriott, Shroud of the Avatar is evidence that the technology can make an immediate practical difference. The claim is ambitious: a game that once carried an enormous persistent backlog can now have no known bugs at the end of each day. Whether that standard holds as the game continues to evolve will be a meaningful test of his confidence in AI-assisted development.

His broader message is unlikely to settle the debate. Instead, it highlights the exact conflict that has made AI so divisive in games: the same tools that may help a small team clear years of technical debt can also be used to justify doing more with fewer people. The challenge for studios will be proving that faster workflows and improved stability do not come at the expense of the people whose craft, judgment, and creativity make games worth playing in the first place.