AI is a Threat to Creativity and Could Kill Game Storylines

Hayao Miyazaki, the creator of iconic anime epics like Princess Mononoke and Spirited Away, once said that AI-generated animation "is an insult to life itself." While I don't think that it's that bad, I do think it's something that game creators and developers should consider when crafting gaming titles using emerging AI technology.

AI has existed in game development on some level for years. While the version of artificial intelligence used in past games often employed rudimentary forms of AI that have nowhere near the computing power of tools like ChatGPT or Midjourney, artificial intelligence, on some level, has been an important part of video game development since the first video games started releasing.

After all, we even have a well-known abbreviation for computer-controlled entities within titles, non-player characters, or NPCs, which need to be driven by something. And while the difference between a logic tree of AND-IF statements and actual generative AI differ in many ways, the distinction doesn't matter in the context of the problems that these tools have solved in the past.

Already, developers have explored using AI for dynamic dialog in adventure games. In games like the recent Darkest Dungeon 2, this opens each playthrough up to new possibilities and paths and helps keep the game fresh by adding a level of randomness.

Darkest Dungeon 2 uses randomly-generated enemy encounters and paths in the game.

While there haven’t been any games that craft an entire story using the power of AI, there are examples of upcoming games and platforms that utilize artificial intelligence to power dialogue and characters in video games. One of these upcoming games, Inworld Origins, features a cast of AI-powered characters that players must interrogate to solve a mystery in a futuristic sci-fi setting powered by the Inworld platform. According to a recent inside look into the game, CEO of Inworld, Ilya Gelfenbeyn, states, “Interactive stories, engaging dialogues, and immersion can be a feature of every experience using Inworld. Work that used to take weeks and months can be condensed into weeks and hours for teams of all sizes.”

Other games, like Starcraft 2, have a computer AI that can beat most average players and have for years. AI-controlled play can even extend beyond video gaming and into other types of games. Famously, all the way back in 1997, Deep Blue, an IBM-powered chess AI, defeated the then-world chess champion, Garry Kasparov.

AI can serve these elements in video games and make the game better for it. It can save development time, help designers solve complex roadblocks, and improve the replayability of many of our favorite titles. However, AI typically fails when trying to replicate humanity, which is something that Miyazaki understood.

AI Isn't Great at Depicting the Human Condition

Gone are the days when video games are relegated to the absolute dregs of media when it comes to general perception. Games like Elden Ring and God of War: Ragnarök, among many others, prove that games can have high artistic merit.

To figure out where AI fails, we can go back to the Miyazaki reaction to seeing AI-generated models crawl along the ground. Relating to a friend who suffers from chronic pain, Miyazaki states, "Whoever creates this stuff has no idea what pain is whatsoever. I am utterly disgusted."

In the video, he remarks how the animated model looks uncanny and inhuman. This same effect can also extend to game storylines, narratives, and dialogue.

Recently, NVIDIA displayed the power of its generative dialogue abilities at Computex 2023. In the showcase, the company demonstrated the power of its model as the player spoke to a ramen shop owner who responded organically using the power of AI.

As John Spitzer, vice president of developer and performance technology at NVIDIA, said: "Generative AI has the potential to revolutionize the interactivity players can have with game characters and dramatically increase immersion in games."

However, regardless of how impressive it all is, there is still something strange about it. Sure, the AI is answering as a human would, but it still feels strangely inhuman, even if it's just an NPC in a theoretical game, and this may be attributed to the uncanny valley effect.

To most humans, the more lifelike and human a robot appears, the more appealing it is—but only until a certain point. Once reaching the point known as the uncanny valley, people often feel unease or strangeness, which results in a negative emotional response. While there hasn't been a term coined for that same feeling when watching a video, speaking to a chatbot, or playing a piece of interactive media, the concept remains the same—the human brain can seem to detect when AI is implemented in certain ways.

Humans Should do the Storytelling

The problem with AI when it comes to human storytelling is that it will always tell the story from a third-person perspective, even if it is written in first-person. AI will always be machine-driven, while humans (at least for the near future) will be human. Part of what makes storytelling good are the feelings and emotions that a game can evoke. This often occurs because of a shared experience and an emotional or spiritual connection one has with the game or its characters.

AI, on the other hand, at least right now, is not sentient enough to understand or evaluate its own experiences in a narrative format because it's just pulling from different human experiences harvested from the web. Just as with digital art, there are serious ethical and moral questions behind utilizing someone else's writing style in games using generative AI. And while I'm a bit biased as a writer myself, I'm by no means anti-AI—I'm just someone that thinks that tech leaders and humanity, in general, should take more consideration when evaluating the limitations of the technology.

The thing about game narrative is that what will often make a game iconic are those instances when the creators speak from personal experience. Regardless of whether you're playing a survivor of a nuclear apocalypse who just emerged from an underground vault, a teen in a zombie-infested post-apocalyptic world, or a cybernetically-enhanced human, these aspects are merely supplements to a world filled with human interaction, emotion, loss, love, and personalities we can all relate to.

For instance, in the popular The Last of Us, the compelling nature of the game isn't necessarily a zombie apocalypse derived from mutated mushrooms. While it is a cool twist on a tried-and-true genre, we've all seen a million different zombie movies, shows, and games. What makes The Last of Us so amazing are the relationships built and broken in the game, the tough decisions players have to make, and the sometimes ugly humanity underneath it all.

In a world filled with ravenous roaming zombies, fellow humans are the most dangerous thing on the planet. If AI generated all the dialogue or narrative, well, I just don't think the game would be the same.

A Call to Invest in Creatives

When big gaming corporations and publishers look at using AI in their games, giant dollar signs flash in front of their eyes. Creatives think of new ideas, innovate, and take risks that AI can't because humans haven't done it first. If we want to live in a hollow world where every game is just maximized for addiction efficiency and sprites and models are plug-and-play using generative AI, I guess we can start using it to create storylines, but the gaming world will be much grayer for it.

What may end up happening, as a result, is that more creatives begin getting pushed out of the space in an already receding industry because corporate executives and shareholders see using AI as cheaper than using human labor. This will create more lifeless games that might give gamers that uncanny valley feeling when playing them because many of the concepts weren't created by a human. And while there will probably (hopefully) always be some level of human oversight on gaming projects, the spirit of games—dialogue, character models, and game mechanics, may be worse for it.

On the flip side, this could all be a doomer view, and we could be quickly approaching the outer limits of AI usage when creating art, at least for now. Gaming companies and developers are better off when using AI in some instances, and it could even lead to more investment being made in the human-generated creative aspects because developers save time by cutting out redundant coding tasks.

Only time will tell how much impact AI will have on the creativity of games. I, for one, hope we keep storytelling to the talented human storytellers at the heart of the games we all love.