Ways to Leverage the AI Revolution for More Efficient Game Development | GI Sprint
The impact of artificial intelligence is evident across all sectors, including the gaming industry. However, while recent AI developments like ChatGPT and Midjourney receive most of the attention, AI has been influencing game development since at least the 1980s.
Tommy Thompson, who runs the consultancy AI and Games, stated during a panel that the idea of AI being a novel threat to jobs is a misconception. He pointed out that machine learning has been part of game development processes since the late 2000s, a continuation of AI practices dating back to the 1950s.
Sean Cooper, technical director at Didimo, shares that AI and automation have always been part of game development, from reducing the need for human testers to designing characters. This long-term use of AI has made development more efficient, especially for smaller studios with limited budgets.
Thompson adds that sophisticated AI tools have traditionally been used in top-tier studios but kept out of the public eye. The recent surge in AI awareness has brought these behind-the-scenes advantages into the spotlight.
In this new era of AI prominence, how can game developers utilize these tools to reduce costs and enhance productivity?
Watch the entire podcast below, download it here, or find it on your preferred podcast service.
Avoiding the AI Fad
The AI surge has resulted in a plethora of tools and services, making it challenging for studios to discern which are genuinely beneficial. The panelists advised sifting through the hype to focus on practicality.
"Conversations around AI in games are often influenced by those outside the gaming industry, often with commercial interests," said Thompson. This has distorted the narrative, affecting both public perception and internal industry views.
"When discussing AI, people ask if certain AI capabilities can be integrated into their games," Thompson mentioned. "While the technology might be used, the legal ramifications concerning copyright cannot be ignored."
"There is no one silver bullet that will create your game for you. I would be surprised if we see that in our lifetime" Lucie Migné
Cooper noted that adding "AI" to a product name can attract investments due to the appeal of new technologies, even if it’s simple automation. This leads to an abundance of AI-related products and services, some of which may not truly be advanced AI.
Moreover, the enthusiasm around generative AI can introduce unexpected challenges and legal issues, notably concerning the unknown origins of training data for these models.
Lucie Migné, senior producer at Mighty Build and Test, highlighted the importance of integrating reliable tools like automation and machine learning into workflows instead of pursuing potentially overrated new technologies.
"They will likely discover, much like we have recently at Keywords [Mighty Build and Test's parent company], that the current tools do not fully deliver on their promises," she asserts. "Learning that can be particularly costly for smaller teams."
Embrace Open Source
With the rapid expansion of AI tools, how can studios without substantial resources create their own solutions? Open source offers a substantial opportunity for such developers, particularly those in the indie gaming space, suggests Cooper.
"Indie games operate on small budgets, but incorporating AI systems—especially free open source ones—can significantly elevate their quality. These AI systems can fill the gaps usually left by the absence of certain hires,” he elaborates.
However, failing to properly integrate and train staff on new tools can have unintended consequences, cautions Migné. The open source community also often points out that the total cost isn’t zero; as IBM notes, considerable running costs must be factored in, and maintaining such software typically requires volunteers to contribute significant time to activities like bug fixes and feature additions.
"Public perception suggests that [generative] AI tools are free to use, but that is misleading," states Migné. "This perception might lead the industry to see these tools as cure-alls for production challenges such as understaffing, missing skills, or experience gaps."
Identify Optimal Workloads for Automation
Determining where to start with automating tasks can be challenging, says Lauren Maslen, Director of Production at Mighty Build and Test, as the industry appears saturated with opportunities for automation. Nevertheless, these opportunities might not always pan out as expected.
She notes that excellent candidates for automation include areas with repetitive content, such as NPCs, characters, and environment duplication, or even coding tasks. At her company, the primary focus has been on optimizing the QA process.
"Consider any game of any scale, where vast amounts of test cases have to be meticulously followed by a distributed team to catch regressions early on. This repetitive nature of the work is a prime target for automation or AI integration,” she observes.
"The challenge is that discussions about AI in gaming are often driven by individuals outside the industry pushing AI products" Tommy Thompson
Thompson emphasizes the importance of identifying mundane and monotonous parts of workflows early on.
"In QA, for instance, a tester might have to run into a door 500 times at various angles to ensure it functions correctly—this can be automated," he explains. "More nuanced tasks, like checking levels for UV errors and texture correctness, might still be better suited for human oversight, despite advancements in image recognition technology."
Strategically Select and Apply Tools
As you explore adopting AI and automation, it's crucial to choose the right tools for specific needs, investing time and resources wisely.
"Not all AI or automation tools are universally applicable," warns Migné. "The task of making a bot collide with objects differs vastly from setting repetitive meeting agendas. It's vital to identify the specific areas for automation and thoroughly research the most suitable tools for those duties."
"There’s no universal solution that will single-handedly create your game. Such a tool doesn’t exist yet, and it’s unlikely we will see it in our lifetimes."
"The industry must be deliberate about what gets automated rather than jumping on trends" Aleena Chia
Dr. Aleena Chia, lecturer at Goldsmiths, University of London, examines how game developers incorporate automation. She highlights that team members may perceive automation differently and stresses the importance of knowing the limits of automation. She uses the distinction between generating individual trees and creating an entire forest to illustrate that AI struggles with broader conceptual tasks.
"Instead of merely following trends, managers and studios need to be intentional about what they choose to automate," Chia asserts.
Remember the Artistic Aspect of Gaming
Ultimately, as studios integrate automation and AI tools, maintaining their unique style and value proposition remains paramount.
Chia advises that automation decisions shouldn’t solely focus on efficiency. Studios must consider the perspectives of all disciplines—from narrative designers to artists to programmers.
Chia anticipates a shifting standard for what’s considered creative enough to remain unautomated versus what’s routine. This evolution will also affect how games are perceived by players, especially those who value the human touch in their gaming experiences.
“It’s hard to predict the exact form games will take, but increased automation and AI integration is inevitable,” she suggests. Chia recounts an artist’s remark at GDC, emphasizing that tasks like texturing, which might seem tedious, actually contribute significantly to storytelling. This underscores the ongoing debate about the limits of automation in creative processes.
