User:Mmcclenahan84/Artificial intelligence art

2020s and generative AI
In the 2020s, text-to-image models, which generate images based on prompts, began to approach the quality of real photographs and human-drawn art. The models used in AI image generation come from large image databases such as JFT-300M which has allowed image generation to continue to develop and become much more accurate and difficult to distinguish from art that was created by humans.

In 2021, using the Transformer models used in GPT-2 and GPT-3, OpenAI developed DALL-E, a text-to-image AI model capable of producing high-quality images based on natural language prompts. DALL-E was one of the first artificial intelligence art platforms to be commercialized and become popular amongst the general public. This model could capture the user's requests and create them with stunning accuracy. Cartoon images capture the child-like innocence, realistic photos appear as if they were natural and could have been taken with a camera, and can capture the style of famous artists with shocking accuracy. Users can also specify that a certain individual or object should appear or not appear in the image, and this model also captures this reliably.

Later in 2021, EleutherAI released the open source VQGAN+CLIP, based on OpenAI's CLIP model.

In 2022, Midjourney was released, followed by the open-source Stable Diffusion, which allowed artificial intelligence art to continue to become more accessible to the general public since the code is available to everyone. Stable Diffusion allowed artificial intelligence art to evolve because it relies on code that will generate a rough image that is close to the prompt given by the user, and then the rough image will go through a process of denoising until the final product is created. Artificial Intelligence art platforms have continued to develop and become more widely accessible due to the open-source nature of the code. This has allowed a plethora of both free and paid software that can develop artificial intelligence art. While many of them are using the same base code as Stable Diffusion, each platform has gone through different evolution and training that allow each platform to be unique. This constant evolution in artificial intelligence has allowed a vast amount of creativity among its users. Artificial intelligence art is no longer limited to just pictures. All forms of artificial intelligence media are being created such as videos, music, and many other forms of media. As these models continue to develop and evolve, they are also becoming extremely realistic and it can be difficult to distinguish between them. This difficulty is beginning to raise the concerns of many individuals worried about the negative impact. Artists are also concerned because there could become an absence of creativity and machines creating the majority of art.

Ethics
AI produced images are causing many artists to be concerned about the way society values them and their output. Artists fear the production of AI will cause traditionally-made art to become devalued, with humans not being able to produce at the scale or price of an AI model.[95]

AI systems have the ability to create deepfake content, which is often viewed as harmful and offensive. The creation of deepfakes poses a risk to individuals who have not consented to it.[95] This mainly refers to revenge porn, where sexually explicit material is disseminated to humiliate or harm another person. This content has predominately targeted celebrities (particularly female) in a manner that is demoralizing and damaging to the individual.

While there is yet to be a method of preventing the creation of deepfake content, legislation has been created to criminalize these actions. Many countries worldwide have implemented legislation that criminalizes revenge porn, and many times deepfakes fall within this category. Another case of this was in 2020, New York Governor Andrew Cuomo created legislation that protects against, " protecting against commercial exploitation or unauthorized use of a dead person’s name, picture, voice, or signature, giving only the families of such people the right to use them." This law also includes deepfake pornography. While the legality of deepfake content is still being debated and questioned, many expect laws of revenge porn to begin including the creation of deepfakes.

A prominent example of this was in late January of 2024, when many deepfaked images of Taylor Swift portrayed in a sexual manner began to flood the internet, predominately on 4chan and Twitter. This brought the issue of deepfaked images to the general public's attention due to the popularity of Swift. Swift also threatened to take legal action against those responsible for creating the images. This incident also drew attention from other celebrities also responding being disgusted that individuals would create these types of images. More individuals were now exposed to the potential threats that artificial intelligence images can present to those depicted. Also, as models continue to grow and become more realistic, it can become hard to distinguish what is real and what is computer-generated. This line continues to blur between what is real and what is generated, and the images of Taylor Swift are just one of many examples of how artificial intelligence can be used in a malicious manner.

The public use of AI art is new, which means laws and policies regarding ownership have not been fully established. This poses a challenge when determining which parties own an image or video.

AI systems gather data in order to create solutions, when gathering data from various sources there becomes the question of whether or not the data can be used to produce a work.[96] Galanter introduces a question of determining how to give credit through the thought process of differentiating the artist and the artistic influences.[97]