Generative A.I. (Concept)

Generative A.I.

CONCEPT

Generative Artificial Intelligence (aka Derivative A.I.) is a type of neural network that can be used to generate a piece of media based on large amounts of training data. These "new" assets are also known as "Synthetic Media" or "AI Slop"

Concept Images Related Pages

EditHistoryPage info

Generative Artificial Intelligence is a collective term for a variety of different artificial neural networks that can generate output from an expansive set of training data. This can be used to stitch together artwork, voice acting, sound, music, text, animation, etc. from its training data into "new" assets. The A.I. part of Generative Artificial Intelligence refers to its usage in the Machine Learning branch of Computer Science. It is not to be confused with other concepts that are called "Artificial Intelligence" in video games such as Artificial Intelligence (Concept) (the programmed behaviour of NPCs in video games) or Artificial Intelligence (a sentient, intelligent lifeform that is made from technology). Generative Artificial Intelligence should also not be confused with Procedural Generation (Concept). While both can be used to create content that is not handmade by a developer, procedural generation does not use neural networks or, usually, machine learning at all. Procedural Generated content in games comes from algorithms programmed by a developer, usually mixing and matching handmade parts together in predictable ways specific to the task it is used for. Generative A.I., however, is trained on large amounts of data and will output based on the input of the prompt and its own data-driven conclusion on what the "correct" output should be given that prompt. Although Generative Artificial Intelligence has been around since the 2000's, its output was usually simple, prone to errors, and restricted by computing power on what type of media it could create (mostly just text and data points). In the latter 2010's and early 2020's, however, new advancements in deep learning architecture lead to neural networks being able handle more training data, learn from that training data faster, and generate more complex assets, like animation, voices, and artwork. Game developers have only dipped their toes in the use of Generative A.I. to create assets in games, mostly because announcing you're going to or being caught doing so has been the source of quite a lot of controversy. Use of Generative A.I. in a game is almost always followed by pushback from a game's playerbase, game developers, and/or the wider gaming community as a whole. A lot of the reasons for this pushback in games echoes the pushback Synthetic Media has seen in other industries:

It removes the possible work for an artist, actor, or other creative individual.
Many neural networks have either admitted to or have been caught using media as training data without the permission of their creator(s), including copyrighted material.
It is seen as a cost-cutting measure that leads to lower quality games and layoffs.
Due to the nature of how Generative A.I. and neural networks as a whole, nothing new is actually created by them. All output is merely reshuffling parts of existing creative work in a different alignment.
Despite advancements in the technology, errors are still common.
Most output is not checked for accuracy or legitimacy, leading to fake information being spread.
It's output can fall into the Uncanny Valley (Concept) for some people, especially with images and animation.
There are very few laws and regulations for Generative A.I.'s use, making it impossible to know if any output wasn't made with stolen work being used as training data.
Developers do not have to disclose the use of Generative A.I. to the consumer on most platforms, which can lead to people being mislead as to what they are buying.
The lack of laws and regulation on Generative A.I. and the already blatant theft of copyrighted materials by many Generative A.I. companies have lead to the prickly situation of not knowing if publishing Synthetic Media is even legal.
It is incredibly taxing environmentally and financially to generate any content using this method in terms of electricity and water usage.

Nevertheless, the video game industry as a whole is fueled by capitalism, worker exploitation, and jumping on tech trends. So even if many people do not want Generative A.I. to be used in game development, big publishers smell jobs they can cut, labour they can steal, and profits they can pocket, so we're unfortunately going to see a lot of AI Slop in the coming years. Note: Games associated with this concept do not need Generative A.I. content to stay in the current build or even a release build of the game. If Generative A.I. was used in the development of the game, the game is associated with Generative A.I. You don't get to cheat off the test of The Plagiarism Machine without doing some plagiarism yourself.

Games view all →]