Prohibitive costs, and expensive vendor lock-ins. Severely limited scalability, and black box systems that prevent customization. Complicated tools that stall creativity, rather than fostering it. These are just some features of the current AI market dominated by a few major players, from hardware providers like NVIDIA to large-language model platforms like OpenAI.

Certainly, some larger enterprises may be fine dealing with these obstacles. But smaller creators and independent artists? Most simply can’t afford to incur the high costs of experimenting with generative AI, plus creating on these centralized platforms can pose serious data privacy and censorship concerns. Worst of all, many contractually state that artists can’t claim copyright over their AI-generated pieces, robbing them of ownership over the works they create.

“The creative process today is quite complex and expensive. The costs can be astronomical, and they inhibit many from doing this type of art at all,” says Malcolm Levy, a co-founder of Refraction DAO, an artist-owned community working at the intersection of digital art, music and culture.

If digital art is to thrive in the age of AI, it must be supported by a system that is fairer to users, one that helps them more affordably create memorable works for the world to enjoy.

A generative AI video created by Jahmel Reynolds using the Regenerate app on Livepeer AI

How decentralization can help AI creators

Using token incentives to support collaboration, so artists can build on networks that are more cost-effective and more accurately represent their beliefs.

Decentralized systems can transfer control of the generative AI tech stack from one centralized entity to a network of individuals working together to create a platform that better reflects their needs and values. This is especially critical to creatives, who may have strong ethical beliefs about how AI should be built, and also may want to more directly have a say over the tools they create with.

Blockchain-based platforms can use cryptoeconomics to bootstrap these decentralized AI communities. By rewarding positive contributions with a token, the network is able to incentivize good behavior and properly credit those who do the most to build and grow it. Creators can choose to align themselves with networks that reward the values they want to see more of, and they can also provide better feedback to improve these decentralized AI tools, since most are open source.

One example of this is Livepeer AI, which is lowering the cost of entry for accessing generative AI. By using its LPT token, it is able to organize and incentivize individual hardware providers to offer compute for a litany of on-demand generative AI tasks, fueling a decentralized pay-per-task model that can be much more cost-effective for AI developers and creators.

Democratizing generative AI art with Regenerate

Blending traditional art techniques and cutting-edge AI technology to shape a more robust, and collaborative, canvas for creators.

Regenerate is an application built with Livepeer AI used for creating new artistic realities with generative AI, including text-to-image services. Built by the Refraction DAO community, it is supported by Livepeer’s decentralized compute network which is already used to transcode millions of hours of video on a weekly basis.

While centralized generative AI tools often have to address many competing audiences with their product, Regenerate is able to focus intently on the specific needs that generative AI creators within its own community face. For example: When surveyed, many of Refraction’s AI artists said they wanted more capabilities for complex collaborative works, a simpler UX/UI experience, and, of course, a more affordable compute landscape.

“The current process can be quite stagnant. To create interesting, creative work that goes beyond just entering a prompt, it takes using a number of applications, a lot of time, and significant cost and knowledge barriers that we’re trying to alleviate,” Levy says.

Regenerate currently integrates AI tools — starting with a simple interface for text-to-image, which will soon expand to include image-to-video — offering a blend of both traditional digital art techniques and cutting-edge AI technology like “shader art,” in which GPU programs originally designed for processing the appearance of 3D models are creatively repurposed to create visually stunning images and animations.

What’s more, it allows AI creators to share and collaborate on generative AI pieces.

"We're using different ways of affecting those tools that are much simpler and easier to understand, so people can augment and create around each other’s work," Levy says.

When artists create AI art on Regenerate, they can check the “This prompt is public” box and then press “Create Prompt” to have their work be shared with the entire community on the app’s dashboard.

Other Regenerate community members can then click on their image and iterate on both the prompt itself or on the different filters and permutations by adjusting various sliders, as shown in the screenshot below:

Empowering AI artists with ownership and community

Improving provenance, feedback loops and community contributions to artistic endeavors, all made programmatic through decentralization.

Refraction hopes to create a model that better serves artists by granting them complete ownership over the works, possibly using blockchain capabilities like on-chain attestation to provide verifiable proof of authorship and provenance. The community also believes it can better respond to feedback from its artists than centralized AI platforms that may make compromises for their leadership teams or investors at the expense of creators.

That focus on collaborative input goes beyond just AI image and video creation, but is woven into the very fabric of the project. Its very code is a work of art for the community, since countless developers from a number of organizations within the Refraction DAO network can creatively contribute to its open source development.

“Decentralization gives the opportunity for all of these different creatives to add on and grow and build AI technology in real time,” Levy says.

By using Livepeer AI, Refraction is able to not just reflect the values of its decentralized community, but also access resources it needs to experiment in the quickly evolving generative AI space.

“There is an ethical value to us using a decentralized compute provider like Livepeer. Plus, there is certain developer help, including tools and functionality, that just wouldn’t be possible to access without their support,” Levy says.

Rather than renting expensive compute space, Regenerate will also benefit from Livepeer AI’s decentralized pay-per-task model, savings that can make the app more cost-effective for its creators than centralized operators.

Embracing the decentralized future of generative AI

Regenerate recently launched on the Livepeer AI testnet, and is actively seeking creators to provide feedback for its text-to-image features.

Learn more about projects building on Livepeer AI, and join the network either as an infrastructure provider or as a developer.