Video games powered by AI and their unsettling future
The use of artificial intelligence by game developers dates back to the 1980s. However, the ghosts pursuing Pac-Man are a long way from digital figures that show self-awareness.
René Descartes seems to have conceived this thought experiment for the twenty-first century.
In a virtual town modeled after “The Matrix” series, residents were being made aware of a bleak reality. Through a microphone, a player told them that they were merely lines of code designed to enhance a virtual environment and that everything else was a ruse. The characters reacted in horrified amazement, strengthened by generative artificial intelligence such as ChatGPT.
One woman wearing a grey sweater questioned, “What does that mean? Am I real or not?”
Two years ago, Replica Studios, an Australian tech firm, unveiled a terrifying demonstration that highlighted both the potential benefits and drawbacks of using artificial intelligence to improve gameplay. In addition to disturbing images in a virtual environment, there is a significant risk. What will happen to the sector as video game companies grow more at ease with outsourcing the work of voice actors, writers, and other professionals to artificial intelligence?
Big tech firms like Google, Microsoft, and Amazon are banking on their A.I. programs to transform the way games are created in the next few years, given the rate at which the technology is advancing.
The technology entrepreneur Kylan Gibbs used the acronym for artificial generalized intelligence, which refers to the point at which computers have the same cognitive capabilities as humans, to state, “Everybody is attempting to race toward A.G.I.” There’s this notion that once you do, you’ll have a monopoly over every other sector.
In the first few months following the introduction of ChatGPT in 2022, discussions on the use of artificial intelligence in gaming focused primarily on how it might help studios create concept art or write simple conversation.
Its usage has grown rapidly. At the Game Developers Conference in San Francisco this spring, thousands of hopeful professionals searching for job prospects were given a chilling peek into the future of video games.
A new program that might eventually replace human play testers with “autonomous agents” that can run through early builds of a game and identify defects was the subject of a presentation by engineers from the artificial intelligence laboratory Google DeepMind.
With a demonstration of how artificial intelligence may analyze a brief film and instantly produce level design and animations that would have taken hundreds of hours to create, Microsoft developers demonstrated adaptive gameplay.
Additionally, the firm behind the online gaming platform Roblox unveiled Cube 3D, a generative AI model capable of creating usable items and landscapes from text descriptions in a matter of seconds.
Following years of significant layoffs, these were not the answers that developers had been looking for; this month’s additional round of cuts in Microsoft’s gaming sector led some observers to believe that the company was investing resources in artificial intelligence.
As the demand for hyperrealistic visuals grew, even the most popular games from studios became financially unprofitable. And some observers are concerned that spending on AI programs in the hopes of reducing overhead expenses may turn out to be a costly diversion from the sector’s efficiency challenges.
In the next five years, most experts predict that artificial intelligence will take over the video game industry, and CEOs have already begun making plans to reorganize their businesses in preparation. One of the first industries to use A.I. programming in the 1980s was the video game industry, with each of the four ghosts that pursue Pac-Man reacting differently to the player’s real-time actions.
Questions concerning the artificial intelligence technology Sony is employing for game creation were not answered.
According to Microsoft representative Yafine Lee, “Game developers will always be at the center of our overall A.I. efforts, and we empower our teams to decide on the use of generative A.I. that best supports their unique goals and vision.”
Beyond what Shigeru Miyamoto, one of Nintendo’s executives, said to The New York Times last year, “There is a lot of talk about A.I., for example,” the company’s representative stated that they had nothing else to say. At that point, everyone starts moving in the same direction, but Nintendo would prefer to move in a different direction.
According to a survey published by the organizers of the Game Developers Conference, generative A.I. has become a widely used tool in the industry over the past year, moving from an idea to reality. Although the majority of survey participants stated that their businesses were employing artificial intelligence, more and more developers voiced worries that it was causing job insecurity and layoffs.
Not every answer was unfavorable. The capacity to employ AI software to do monotonous jobs like positioning barrels throughout a virtual town was hailed by certain developers.
Even though the conference in late March included some amazing technology demonstrations, several developers acknowledged that their applications were still years away from being widely used.
“The gap between prototypes and production is enormous,” stated Gibbs, who leads Inworld AI, a technology business that creates artificial intelligence solutions for retail applications in industries like gaming, healthcare, and education. For Microsoft, he participated in a conference panel where the corporation demonstrated its adaptive gameplay model.
Large studios may have to pay millions of dollars to update their technology, according to Gibbs. By offering A.I. tools that would compel studios to join their servers under pricey agreements, Google, Microsoft, and Amazon are all vying to be the next backbone of the gaming business.
The team that produced the tech demo based on the “Matrix” franchise, Replica Studios, has surpassed the rapid advancement of AI technology. Because of the speed of competition from larger firms like OpenAI, Replica closed down this year.
According to Eoin McCarthy, chief technology officer at Replica, the platform was at its most popular when users were creating over 100,000 lines of conversation from non-player characters, or NPCs, which cost the business around $1,000 each day to run.
He noted that while the cost has decreased in recent years as the A.I. programs have gotten better, the majority of developers are unfamiliar with these limitless expenditures. There were also concerns regarding the potential cost of NPCs communicating with one another.
Some gamers began to worry about the future of the NPCs when Replica announced the end of the demo. “Were they going to continue to live or would they die?” McCarthy recalled players asking. His response would be, “It’s a tech demonstration. These individuals are fictitious.”
In their presentations to studio executives, big businesses sometimes ignore those ethical considerations.
To give the NPCs in a cyberpunk ramen restaurant real-time conversations, Nvidia has worked with a company called Convai. A video on The Verge showed that Sony was using OpenAI’s speech recognition technology and other tools to replicate Aloy, the main character in Horizon Forbidden West, in a way that allowed her to respond to questions from players.
Some engineers have gone even farther by testing AI applications that insert realistic representations of real individuals into video games. Researchers from Google and Stanford University collaborated on the development of generative agents in late 2023, which they defined as surrogates for human behavior.
Their report stated, “Generative agents wake up, prepare breakfast, and go to work; artists paint, while writers write; they develop opinions, observe one another, and start conversations; they reminisce about days gone by as they plan for the following day.” These agents formed friendships with one another in a virtual environment based on The Sims organizing a Valentines Day party at a café.
The advancement of technology that may reduce the strain of obtaining human test subjects has been praised by some ethics professionals. However, others have questioned the value of a technology that can just mimic a person’s choices.
Celia Hodent, a cognitive science and user experience expert who has been working on a code of ethics for the video game business, said, “Humans should be at the center of what we do.” “Instead of thinking of A.I. as a solution for everything, having better processes might be a better starting point.”
Several of the existing software that may automate game creation are still very buggy and excessively priced to use. Entrepreneurs are advising patience, arguing that it will likely take another five years for practical models to advance in quality and affordability.
The adaptive gameplay model demonstrated during the Microsoft conference session would likely cost hundreds or thousands of dollars per hour to operate commercially, according to Gibbs. He added that Oasis, a similar program, has its own issues. It forgets visual data not immediately visible on the screen since it creates material frame by frame, leaving users in a dynamic world.
Gibbs said that although the technology holds promise, it is still a solution looking for a problem.
“Is it really less expensive if it costs you 5,000 times as much to operate a game, even if it’s a cheaper method to create games?” he questioned. “How do we move the research community in a more helpful direction?”
Ethics specialists are still concerned about the industry’s readiness for sentient creatures and self-designing levels, regardless of the dollar signs.
Normalizing the technology would jeopardize individual agency and privacy, according to Cansu Canca, the director of responsible AI practice at University in Boston.
“My greatest worry is not whether the A.I. develops consciousness,” she stated, “but rather what it means for our existence in a virtual world where encounters cannot always be regulated or predicted.”
