How AI Is Replacing Game Developers

For example, the author of Terraria at the beginning of 2023 dedicated a series of tweets to his dialogues with the neural network. In one of his messages , he provided a screenshot of the neural network’s response to a request: “ Act as if you were Terraria 2, a text adventure game. I will type the commands, and you will print the responses that should appear in the terminal. Output responses in the terminal formatted as program code. The player starts the game with a Copper Gladius, a Copper Hoe and a Copper Axe. There is a Guide and a green slime nearby. Some time passes with each team. Start the game with introductory words and describe what surrounds the player .”

To which the AI ​​gave the following response:

Welcome to Terraria 2!

You find yourself in a dense green forest. The rays of the sun break through the branches of the trees, and a light breeze sways the leaves. Standing next to you is a friendly person who introduces himself as the Guide. He says, “You have just arrived in a strange new world, full of many dangers and treasures. You just need to find them.”

You look around and notice a green slime not far from you. He appears to be harmless, but the Guide urges you to be careful when exploring.

You begin your adventure with a Copper Gladius, a Copper Hoe and a Copper Axe. Type “Help” to get a list of available commands.

Subsequently, Andrew Spinks entered various commands and the neural network responded to them adequately. In addition, the AI ​​gave advice on what to add to Terraria, suggested a name for a unique artifact with an explanation of why such a name suits it, and told a thematic joke: “Why does a creeper come to the Terraria server? To blow up the game! ( Why did the creeper join the Terraria Server? Becouse he wanted to blow up the competition “.

Perhaps the answers and advice of the neural network will seem unoriginal or banal. But if you look at what plots develop in 99% of games outside of the top AAA titles, then the scripts written by a neural network will turn out to be at least no worse. And if the money and time saved on screenwriters are spent on an editor, they will most likely turn out to be even better.

However, the neural network is not yet capable of replacing natural talent among writers or screenwriters. So Neil Druckman can rest easy for a while. But the neural network will most likely take on some of the secondary work for the foreseeable future.

Thus, the enthusiast integrated the ChatGPT chat bot into the game Mount and Blade 2 Bannerlord – this made it possible to conduct full-fledged dialogues with non-player characters (NPCs). The player did not select the response options proposed by the authors, but entered text messages and the neural network responded to them adequately, without leaving the role assigned to the NPC. To ensure that the answers fit within the given restrictions, the neural network used data about the game, factions, professions, upcoming events and locations.

We can only regret that such a tool was not in the hands of the developers of Fallout or Mass Effect!

However, one nuance is still worth noting. Modern neural networks behave like a gullible child and take everyone’s word for it. This causes confusion. Having listened to enough active online trolls and city crazy people, the AI ​​begins to produce such gems that people immediately shut their mouths. Thus, a seemingly harmless anime girl, as well as an AI streamer, received a ban on the Twitch service for “inciting hatred.”

So neural network writers cannot yet do without an adequate human teacher.

AI as an artist

While neural network scriptwriters remain in the shadows for the time being, the fame of neural network artists resounds throughout the Internet. Midjourney gets a lot of praise. This AI creates amazing images based on text descriptions.

Some “humans” decided to rebel due to fears that artificial intelligence would deprive them of their jobs. The fears are not unfounded, but this is not the first time such a situation has arisen in the development process.

At the present time, only some reenactors will be able to create a stone ax, just as today they are not involved in smelting swamp ore. Technical development in the 20th century, even without computers, buried many professions (for example, a chimney sweep or an elevator operator).

The massive development of computer technology has deprived typists and computer workers of their jobs . So ordinary artists here, alas, are not first on the list. If your work is at the level of AI creations, then no matter how offensive it may be, you will have to either become better or look for another source of existence.

Today, AI artists are used as assistants in the gaming industry. They create sketches based on text descriptions and are sometimes able to inspire fresh and interesting ideas. It’s a damn handy tool. But he is not yet able to replace the artist’s work.

It is doubtful that a neural network can create an artistic style without explicit human prompting. And it’s unlikely that a neural network from scratch will be able to give players the kind of world that the creators of Atomic Heart imagined. But AI is capable of developing the ideas of artists and designers in a certain direction and pushing them to create something new.

What AI is really good at is helping with image restoration and enhancement. Thus, during the recent presentation of the 40th series, NVidia demonstrated the work of neural networks to improve game performance (DLSS 3.0 technology). Artificial intelligence upscales images to 4K resolution without any visible loss of quality.

In addition, Nvidia showed the capabilities of AI to improve the graphics of older games. The created technology allows you to create remasters of games with one click of the mouse. The graphics updated in this way look better than in old games, but they do not reach full-fledged remasters and remakes.

A striking example of how not to use artificial intelligence to improve graphics was demonstrated by Rockstar. The company released GTA The Trilogy Edition with AI-enhanced graphics, which caused a storm of negative emotions among fans.

There are questions about the use of AI to create character appearances and to Blizzard. The company uses procedural generation of appearances with their subsequent refinement by artists. But the artists at Blizzard are not as attentive to detail as the fans of the game. Neural networks are not yet able to correctly display some details and in such moments they need human help. But if a person is lazy or inattentive, then “jambs” will creep into the release.

Therefore, it is premature to talk about replacing artists with artificial intelligence. AI can replace artisans, but not talented creators. Neural networks make it possible to improve old games and create new ones, but they do not replace people. Left without proper supervision, a neural network can update graphics in the spirit of the updated GTA The Trilogy. A person is required to be a mentor for AI, and only joint work can produce results.

AI as a game designer

Game designer is a key position when creating games. This person not only manages the process of creating the game, down to the nuances – he is responsible for the basis, for the game mechanics and their configuration.

Artificial intelligence has not yet shown itself in this area and, probably, will not be able to show itself in the near future. How to train a neural network to bring people joy from games? Everything is clear with the visual and textual components, but creating new game mechanics is much more difficult.

But if you look at the world through the eyes of a cynic, the future of AI in game design will not be so rosy. Companies are already using AI to improve the efficiency of monetization of their projects and to control the playing style of players .

AI monitors chats for toxicity, monitors players and the inadmissibility of certain actions in the game. AI changes the difficulty of games not only to ensure that the player enjoys the game, but also to encourage him to make in-game purchases. Neural networks, without being game designers, are already influencing gameplay.

AI can be a good assistant for a game designer. Artificial intelligence will take over routine operations, such as creating sketches based on text descriptions, creating dialogues, some descriptions, and, ultimately, will push a person to new ideas. The time saved on routine operations will allow an enthusiastic and competent person to develop them. Who knows what games we’ll see in the future.

But history could take a different path. From being an assistant to a game designer, AI can turn into an assistant to an effective manager.

AI as an effective manager

AI will be able to free not only game designers from routine, but also management. The development of artificial intelligence will eventually lead to neural networks running companies and taking over some of the routine work. Perhaps this would be a way for Blizzard or Ubisoft to get out of the deep hole in which these companies find themselves, but the AI ​​boss is still a fantastic creature.

AI as an effective manager is still fantastic. But time goes by. For us, for the players, it will be important how the effectiveness of the AI ​​manager will be measured. I would like to see more satisfied players. But let’s face it: most likely, effectiveness will be measured by the amount of money earned from players.

Neural networks are an extremely effective tool in the right hands. Like any tool, it can be used for both good and evil. Today, no neural network can completely replace a person when creating games. Perhaps in the future, some of them will be able to create games using only text descriptions, just like they create images. But someone has to create this text description and the name of this person is a game designer.

It is the game designer who is responsible for the game from beginning to end. Not artists, composers, programmers, managers, etc. The vision of a game designer determines everything from game mechanics to visual appearance and musical accompaniment. In working on these components, AI will be a very good assistant for a talented person, although the participation of talented people will also be useful. If such a tool ends up in the hands of a passionate team of like-minded people who can create games whose gameplay will captivate players and give them joy, then progress in the field of AI can only be welcomed.

But so far it seems that such a goal is not set for the creators of AI. Game company executives are interested in reducing costs (AI works as a screenwriter, artist, composer, programmer) and increasing profits (AI optimizes monetization, moderates chats and fights toxicity, and adjusts the difficulty level). As always, they forgot to ask us whether the players need such AI.