We recently featured Christoffer Holmgård, CEO of modl.ai, in our new AI Gamechangers newsletter. This is a follow-up opinion piece where he writes about how AI-powered bots can transform the industry by making expensive and reputation-damaging buggy game launches obsolete, as demonstrated by the contrasting examples of Cyberpunk 2077's problematic release versus Warhammer 40,000: Space Marine 2's successful launch.
As AAA games continue to grow in scale and complexity, buggy launches have become the norm.
The issue-ridden release trend is driven by market dynamics that prioritize speed over quality, but the benefits of winning the race to market do not outweigh the costs of rushing out a sub-par, bug-ridden product. From financial losses to reputational damage and delayed release cycles to stressed-out devs, the fallout from forcing post-launch bug fixes is immense.
Take Cyberpunk 2077 as a case in point. Following the infamous 2020 release, gamers ripped Cyberpunk to shreds online. CD Projekt Red (CDPR) issued refunds. Sony removed the game from the PlayStation store. CDPR spent $40 million on post-release bug fixes, updates, and expanded console compatibility. And in an attempt to win back gamers, the studio also spent over $80 million to develop and market the Phantom Liberty game expansion.
The $120 million total investment was intended to salvage Cyberpunk's chance of long-term success and restore its reputation. After seven years of building anticipation and garnering accolades from the community, CDPR flushed all that goodwill away with a buggy launch. Why didn’t the studio take just a little longer to identify and address the catastrophic glitches before launching?
Developers don’t intentionally release flawed products; rather, these kinds of buggy game launches are a byproduct of economic pressures. But to ensure bug-free releases without slowing down development cycles, AAA studios (and developers at all levels, for that matter) need advanced AI tools for game development, including AI-powered game testing bots like those found in modl:test.
By dramatically reducing the time needed to identify and fix bugs before they make it into production, AI game-testing bots are poised to revolutionize game development. These tools promise to make buggy launches a relic of the past, and are quickly becoming a competitive differentiator for major game studios. Developers seeking to remain competitive and achieve stable, high-quality releases will be under the gun to participate in the arms race for AI-powered QA solutions.
Production bugs are considerably more expensive to fix than development bugs
CDPR’s Cyberpunk woes were a highly visible example of the buggy release trend. Fallout from glitchy launches isn’t limited to AAA developers either. Some expert developers estimate that once the game is released, fixing a single bug can cost game studios $50,000. Fixing a bug post-release is more expensive than fixing a bug in production development because of the interrelated nature of game design — and of modern business.
Imagine you’re installing a wingnut on a commercial airplane. If you find out the wingnut is broken once it’s in place, you have to ground the plane and get it back to the maintenance hangar. Then, you have to replace the faulty wingnut with a functional one. Next, you have to run holistic tests to make sure the new wingnut is working properly and see if the faulty one caused any secondary issues. You also have to check that no other damaged parts were installed elsewhere in the plane, and that no other planes were using the faulty wingnut.
All the while, all impacted planes are out of commission, impacting flight schedules and stressing the rest of your fleet. Customers are complaining on social media, and your airline’s reputation is taking a beating. If you had caught the broken wingnut before installing it on the plane, fixing the problem would have been as simple as tightening the bolt. You would have avoided any knock-on effects to the rest of the airplane and your business interests. As Boeing knows, all it takes is one faulty door.
Although lives aren’t at stake with games the way they are with airplanes, the metaphor holds true. Complexity increases exponentially in an operating environment compared to on the factory floor. The far-reaching implications of production bugs span decreased playability and negative public sentiment to PR crises that permanently set studios back. And that’s before addressing the financial costs associated with actually fixing the bug.
Advancements in AI and gaming technology have made post-launch bugs unnecessary
Some argue that it’s precisely gaming’s complexity that necessitates human input. The highly interrelated nature of game development means that a bug can fan out across multiple areas of specialization, like programming, graphics, and audio. Because so many development areas converge to create the game experience, ascertaining the nature of a bug must require human intelligence.
That may have been true a decade ago. When we started Modl.ai, in-house automated testing procedures required so much maintenance that the systems weren’t worth it. Back then, automating testing meant configuring bots to perform specific actions, which were immediately invalidated when the game changed. Achieving specificity meant building either brittle systems or subpar solutions.
But the scripted bots of the past can’t hold a candle to today’s AI-powered QA tools. Since 2018, developments in deep learning and search techniques have unlocked a viable technical advantage. Transformer systems like large language and visual models have augmented those outcomes by orders of magnitude. Together with the recent significant increase in compute capabilities, all these developments have sparked a clear inflection point for what’s possible with AI.
At the same time, the games industry has become more modular than ever before. In the old days, developers couldn’t trust they’d be able to consistently achieve their creative goals using someone else’s engine. But third-party tools and components have gained popularity because they enable standardized development at scale. Sure, some AAA studios build and maintain proprietary game engines today. But more often than not, major publishers use third-party systems like Unreal Engine. Building game engines from scratch and hunting for bugs manually are no longer effective uses of developer time.
The future of games development in a market that won’t tolerate buggy releases
Because bigger games mean more and harder to find bugs, staffing sprawling QA teams is simply not an economical or cost-efficient option. Instead, the continued evolution of AI and games technologies will catalyze the market-wide shift away from buggy launches. Instead, armed with automated QA bots, well-tested games that run smoothly on launch day will once again become the norm.
Practically speaking, that means Cyberpunk-style fiascos will pale in the shadow of successful hygienic launches, like Warhammer 40,000: Space Marine 2. A surge in purchases skyrocketed the title to #1 on the charts in September 2024, and overwhelmingly positive reviews have kept that momentum going. Gamers rewarded the bug-free launch with commercial success and a clear path to ongoing growth. Saber Interactive even published a year-long post-launch content roadmap before launching Space Marine 2.
Developers who don’t invest in the tech solutions that allow them to satisfy gamers’ increasingly high standards at scale will be outpaced by those who do. In that sense, automating testing with AI-powered tools is a de-risking strategy that ensures cost-efficiency and competitiveness. Since one of the most important elements of a game’s competitiveness is time to market, automating QA testing is critical. Taking the latency out of testing makes entire development teams move faster, and is the best way to win back development time and shorten iteration loops running at scale.
Thinking any tech solution will afford your team infinite development loops is a mistake; studios should take care to set reasonable project scopes and prioritize their efforts strategically. But investing in automation tools that increase the density of development cycles on committed features and support successful bug-free launches will always be worthwhile.
Thanks to modl.ai's Christoffer Holmgård for submitting this piece. For more weekly interviews about AI in games, follow the AI Gamechangers newsletter on Substack.