October 17, 2022 AI and Art: 5 Tech Projects that will Change Art Forever

By Sam Ancer

Art and AI

Artificial Intelligence, or AI, refers to any computer that is capable of performing a role normally demarcated for a human. Essentially any time a process is automated by a machine, that was once held by a human, that can be seen as artificial intelligence. Recently AI has been making waves in the art community. This is because there are a number of new AI projects that have been involved in automating artworks. 

Critics are often concerned with how AI uses existing art as reference without providing credit to the original creator. There are also concerns around the future of art, if AI is able to create work at a greater scale and pace than human artists, what space would there be for human creators. People who champion AI in art point out that reference is something already done by human artists and is an important part of how art evolves over time. Champions of AI and art also view AI as a tool to help artists, rather than seeing it as a way to replace the human element of art. We’ve put together some of the most exciting AI art projects across mediums in our article below.

Dall-E

Dall-E is an image generator that uses text captions to create artworks based on referenced images on the internet. What this means is that the Dall-E AI will attach specific images to specific keywords and use those as a reference to create an image. For example, if you provide the prompt “A dog sitting in a chair” then the Dall-E AI will collect all of the images of a dog, a chair, and sitting, and use that to create a picture. 

Dall-E is also able to adjust to different colours, perspectives, and art styles. The AI has recently had an upgrade, with Dall-E 2 being able to create higher quality images. Dall-E 2 is also able to retouch pictures with new items. It also operates faster than the first version.

Dall-E 2 is still under research as the AI image generation software is still trying to remove the possibility of creating offensive imagery. This is great because concerns around AI and reinforcing biases and bigotry have been at the core of the conversation around AI and art for a while. These problems can happen because of offensive content that exists online. When you’re dealing with billions of sourced images it's not possible to filter out every example of offensive content manually.

The success of Dall-E has led to the creation of many other text-prompt AI image generators. We could soon see the creation of an entirely new medium of art, and by extension artist. For those that are worried about what this means for the future of artistic expression, well, it could mean that the next best artists are not those with technical artistic skills, but those who know how best to interact with AI and get the perfect image.

This is a good thing. While artistic skill and technique is something that should always be admired, it is not something that is accessible to everyone. Some people lack the formal training, others may be unable to physically paint due to disability, but we all have creative minds and a desire to create. Text-prompt image generators should be seen as a tool to empower those that, for whatever reason, are unable to participate in classical art processes.

AIVA

In the movie “Her” the AI software, “Samantha” creates a piece of music as a way to capture her relationship with the movie’s protagonist Theodore. AIVA founder Pierre Barreau was inspired by this scene and the idea of using music as a way to capture moments in our lives. AIVA stands for Artificial Intelligence Virtual Artist and uses neural network machine learning to compose music.

The goal of AIVA is to create personalised soundtracks that can capture moments within our lives. This would be impossible to do manually, as the creation of a soundtrack in and of itself takes many months if not years. In order to scale this process an automated system like AIVA is needed. 

Using over 30 000 musical scores as its database, AIVA is able to compose songs based on genre, mood, and era.

AIVA allows its premium users to maintain all copyright for everything they create for the platform. 

Copymatic

Copymatic is an AI copy generator that uses a few text inputs to create copy based on whatever needs you might have.

It has around 50 different types of generator to suit varying copy needs. From hook generation to entire blog posts, the AI just needs a small amount of input to create any kind of copy you might need.

While some people may view the increasing role of AI in writing as a threat to human jobs, others see it as an opportunity to create new kinds of jobs for people who are comfortable working with technology. Ultimately, AI is likely to have a major impact on the way we write and consume text in the years to come.

The Copymatic AI is quite good, so good in fact you might not have realised that the above paragraph was written by the Copymatic AI.

Concerns around this kind of generated copy content is that it will ultimately remove the creative element of copywriting. The personal touch and flair created by a human being will be ironed flat in favour of a streamlined AI generated text.

However, much like text-prompt images, the skill and flair would come in the form of the way the user works with the generated text, what changes they make to what’s been created. Much like other forms of AI art, there is always space for the human element.

Games by Angelina

AI has been a part of video games since the 1950’s with the creation of Nim.

Now there is an AI that is creating entire games itself. Games by Angelina are a collection of games created by an Artificial Intelligence. The AI itself is created by AI researcher Mike Cook for his Phd.

Angelina works through evolutionary computation. It uses a trial and error system to solve problems, creating various “breeds” of a solution and allowing the successful “breeds” to evolve and mutate to come up with new solutions.

Angelina is more focused on game design than mechanic creation and so often borrows from third party libraries to create the mechanics it uses for its games.

It’s unlikely that Angelina will disrupt the games industry by itself, being too small in scale and scope. But the idea of AI generated game design is an interesting idea.

Games are complex, and involve a variety of different art forms to work together in order to create a single project. From the design of game mechanics to story writing, music scores and the creation of art assets, a virtual game designer would have a lot to create in order to be a fully automated process. This doesnt even factor in other elements like level design, balancing, pacing, and all the nitty gritty pieces of game development.

But the idea of AI game design is exciting, and Angelina can be seen as the origin for this possible new avenue of games.

Imagen Video

Google’s Imagen Video is a text-prompt video generator.

The AI generator is able to create videos around 5 seconds long. It uses an image library of over 60 million images to help create the individual frames that will make up a video. 

Similar to Dall-E, the video generator uses prompts connected to keywords that calls up images and creates full scenes based on the prompts given. 

Google plans on working with Phenaki, another text-prompt video generator to create longer, if lower quality videos.

While we are a distance away from creating full length feature films using AI, the advancement of Google’s Imagen offers us an exciting look at what AI generated video content could become. With things like deepfakes already becoming more capable over time, we could be entering into a world where all one needs to create a movie is a few text prompts and some creative thinking.

Art and AI

While we will likely never fully replace the human element of creative works, and we should never want to do that, we can see how AI might empower the creation of art. We can remove the existing barriers to creative expression and allow for everyone to have a voice.