Will AI revolutionise or ruin video games?

"That's ultimately the utopia of AI that I imagine: Everyone's just more creative. We'll have more beautiful digital experiences. People will become closer together through these shared creative experiences," says Sasha MacKinnon, founder and managing director of Mino Games. "I think it's possible, we just got to do it right," adds the Australian, a self-confessed "Pokemon nerd" who started developing games at the age of six.

MacKinnon's enthusiasm is contagious; he beams as he talks about the opportunities that lie in the use of artificial intelligence — AI for short — in game development.

Many other gaming insiders have come to a similar conclusion. According to a study by management consultancy Bain & Company, published in September 2023, the majority of the gaming industry executives surveyed believe that AI will be involved in half of the development process in the next five to 10 years.

AI creates 'magical moments'

For example, in the role-playing game "Dimensionals," developed by Sasha MacKinnon's studio, AI is built into a particular character who is a slightly crazy nice guy, brimming with creativity, who always joins the others at the table and keeps making one funny suggestion after the other.

AI can generate characters, fights and dialogues: "We're using AI to have these magical moments where the characters just say something that you totally didn't expect that really isn't scripted and is based on your history of playing with them."

For now, AI quickly reaches its limits: 40-50% of its suggestions are unusable, says MacKinnon. So at Mino Games, AI serves as creative collaborator who provides ideas that are then implemented by humans.

The sketches of the monsters for "Dimensionals" were generated by AI, but the monsters that appear in the game were created by human designers.

Game developers don't have to worry about their jobs, says MacKinnon. They cannot be replaced by AI, he believes, adding that "games are the purest form of human expression," as they not only combine story, art and sound, they also offer "an experience for people. AI will never, ever be able to do that as well."

AI exacerbates precarious working conditions

Daniel Landes does not share this optimism. He works as a freelance translator and adapts international games into German, including "Hogwarts Legacy" and "Overwatch 2."

Translators usually work as freelancers and are paid per word. Nobody gets rich from it. But now, as the use of AI is being required by game producers, they are also expecting translators to be paid less for their work — only 30% of the original per word price.

Translators would like to use the AI tools as an aid, like mathematicians use a calculator, explains Landes. "The main problem is that we are forced to work for less money, even though it is not proven that work will be made easier." The more complex a text is, the worse the computer translation, requiring more editing work. "The machine doesn't understand context, but rather calculates probabilities," says Landes. "Especially with dialogues or creative texts, it's not about translating literally, but about capturing the meaning of the original. AI cannot do that."

The umbrella organisation for audiovisual translators in Europe, AVTE, published a manifesto on AI translations in 2021, in which Daniel Landes was also involved. The authors warn of a loss of quality, question the efficiency of the tools and demand fair pay for translators.

However, there has been no improvement in the general conditions so far. "Because most translators are freelancers, it is incredibly difficult to organize as a profession. Many translators have to struggle very much to even make ends meet."

It's not just in Germany that the profession is facing difficult times. "All translators worldwide are under pressure," says Landes. "We see the same problems everywhere: Our names are not mentioned in the credits, our fees keep being lowered and we have no say or right to negotiate."

AI won't automatically get better

AI is only as good as the data it is fed with. In addition, it is by no means certain that AI will improve over time. It can also get worse, as researchers have found using ChatGPT as an example.

It is therefore not unreasonable to think that the quality of games will be affected if AI is increasingly used.

But since the technology is often faster and significantly cheaper than human translators, game designers or copywriters, it could still prevail and lead to people losing their jobs and studios simply accepting lower quality as the lesser evil.

"One of the biggest risks of AI is that people become complacent," says game developer MacKinnon. "And this is one of the big things about AI in general: We're at these crossroads. You know, it could be the best thing that happens or it could really damage humanity."

We should not blindly rely on AI, he points out, but rather use it as a tool, for example, for inspiration. And it should encourage people to develop their own stories and games. He expects a flood of new games from people who have never developed a game before, but are now able to do so thanks to the technical possibilities.

In fact, the most successful games in recent years have come about because existing games were modified and further developed by the community — i.e. by the players themselves, not by the game industry.

Among others, "Counter-Strike" emerged from a modification of the first-person shooter game "Half-Life."

"Fortnite" is a variant of "PlayerUnknown's Battlegrounds," whose battle royale concept — 100 players compete against each other and the last survivor wins — was invented by a so-called "modder," a term used to describe those who redesign games by creating new levels and maps.

AI makes game development more accessible

One possible scenario is that using AI in game development will create more games, but the quality will drop, which means that more boring, mediocre games will be released.

On average, more than 30 new games are already appearing on the Steam gaming platform every day.

Additionally, AI — which can handle all tasks of game development, from graphics to copywriting to programming — makes it much easier for everyone to develop a game.

This means that more user-generated content could emerge in the coming years, which may also create new, successful genres.

After all, the development of technology — which was initially rather limited — has always had a significant influence on game development, starting with text adventures in the 1980s, followed by 2D platforms with pixel graphics in the 1990s, then detailed open-worlds games in the 2000s, up to the currently popular multiplayer online games, which bring together dozens of players at the same time.