At the Intersection of Generative AI and Blockchain: Tokenizing Creativity

We are continuing our deep dive into the convergence of AI and blockchain technology with an in-depth look at Generative AI and the role of tokenization. As one of the most innovative yet controversial areas within AI and blockchain technology, we will need to pull from our earlier articles on RWA tokenization, NLP in AI, and AI’s role in risk mitigation and cross-chain interoperability to fully grasp the broader consequences of the merging the two technologies.

In this article, we will examine the power of Generative AI, how it functions, and the issue of copyrighted materials when monetizing content from Generative AI outputs. We will then turn to the role of blockchain technology and the use of NFTs as a potential solution to these issues. We will also examine industries already leveraging NFTs before closing with the potential future for this dynamic space and what role AI might play.

Generating the Future of Content

Similar to what we have seen in the rest of the AI space, the evolution of generative AI has deep roots in computer science that date back to the 1960s. British artist Harold Cohen achieved the earliest iterations of computer image generation at the University of California San Diego with his AARON project. Yet despite these early iterations of generative AI imaging, it would not be until the modern AI boom with the introduction of ChatGPT3.5 at the end of Q4 of 2022 that the broader public would get exposure to this once-in-a-generation technology.

GenAI and prompt engineering were swiftly inserted into the public lexicon along with LLMs when generative image protocols exploded in popularity with the release of Midjourney, Leonardo.ai, and DALL-E in 2023. Overnight, everyone was given the power to generate photo-realistic images in seconds that had previously been labour intensive and reserved for professional artists and photographers.

Since then, generative AI has grown by leaps and bounds, continually iterating and improving on the earlier versions of itself. Even legacy Web2 companies began implementing versions of AI image generation and editing protocols, like when Photoshop introduced its generative fill toolkit in May 2023. We have also witnessed the sector's expansion beyond images into audio, video, and 3D modeling.

How exactly does Generative AI work? Do traditional artists need to worry, and where can blockchain assist Generative AI?

Understanding the Technology

To framet where blockchain may intersect with Generative AI, we need to first understand how the technology works and how it may or may not be interpreted as plagiarism.

The first step in Generative AI is the same as in other AI models—the collecting, indexing, and cleaning of raw data. Generative AI collects images, audio samples, video, or 3D digital models. From there, the model can be trained to identify objects, textures, colours, and audio patterns. 

Once the model has broken down its sample data into its most basic components, it can then be used to reconstruct and replicate patterns and dependencies like how colours interact and the spatial relationship between objects. Similar to how an LLM would use probabilistic models to predict the next word, sentence, or paragraph, Generative AI uses probabilistic models to predict pixel values and locations in relation to each other, combining them into a single cohesive image output.

The final stage of Generative AI is to utilize these outputs in its feedback loop. Iterating and improving the models to create more precise outputs over time.

‍

The controversy over copyright starts to blur because the models can get trained on open-source data and do not directly copy any individual piece of raw data for replication. They use highly sophisticated prediction models based on billions of fragments of raw data touchpoints and combine them into a single output through predictive modeling. On way to think about is that the models are more akin to how a modern singer may be influenced or inspired by Michael Jackson or the Beatles rather than directly covering one of their songs.

The Rise of NFTs

Non-Fungible Tokens first arose in 2014, when digital artists Jennifer and Kevin McCoy first minted Quantum on the Namecoin blockchain. NFTs would begin to gain a niche following within the space in 2017 with the release of CryptoKitties and later go viral in the 2021 bull run along with projects like the Bored Ape Yacht Club, CryptoPunks, and independent digital artists such as Beeple. 

During the 2021 bull run, NFTs demonstrated the power of the underlying blockchain technologies use case. The immutable decentralized ledger could improve the long-standing issue of establishing coherent provenance. By having a permanent and unchangeable digital stamp of authentication, industries could easily identify legal ownership over their products. High-end art database Artory has excelled in utilizing blockchain technology to establish provenance over exclusive art pieces.

‍

Since the height of the 2021 NFT boom, NFTs have retreated in popularity but remain no less important. Introducing dynamic and semi-fungible NFT projects through ERC-721 and ERC-1155 token standards has created new markets with the rise in real-world assets (RWAs). The tokenization of RWAs, specifically in real estate and the automotive industries, has benefited from the ability to establish cohesive provenance while updating the NFT with maintenance and improvements over time.

Minting NFTs

NFTs gained popularity during the 2021 bull run because of the ease of minting an NFT collection. For a hypergrowth industry that was relatively niche with technological barriers to entry, the ability to mint NFTs on platforms like OpenSea and Rarible enabled an easy entry point and helped onboard millions of users. It could be more difficult to navigate a wallet setup than creating your own NFT collection. 

‍

The initial setup was done through a simple account creation process. From there, once the user had connected a wallet to their account, they could easily upload and mint a collection within minutes, similar to the ease of uploading their images to their cloud provider. The user experience was unparalleled, and once their images had been approved by moderation, they could easily transact in and out of the platform to their exchange of choice.

The Fluidity of Digital Art

The ability to mint NFTs and freely transact in and out of digital art was an important step in onboarding millions of people. It was a crash course on the market volatility of the crypto space, but more importantly, it served as a dynamic education tool for users. They very quickly understood and began implementing crypto transactions. For example, seamlessly moving from the NFT platform to wallets and exchanges and back into fiat.

It also allowed many creators to monetize their digital art for the first time. It represented the underlying promise of Web3, putting financial and creative sovereignty back into the hands of the individuals rather than third-party gatekeepers.

A New Age of Royalties

One of the often overlooked aspects of NFTs, when it came to establishing provenance over digital assets, was the ability to layer in automated royalty payments to the original creator. While this concept of Artist Resale Rights (ARR) or droit de suite has existed since the early twentieth century when France first introduced it in 1920, it remains a newer practice for many states.

NFTs provide a unique opportunity in this regard. The automated process of implementing royalties on any transaction with a specific NFT addresses this issue without the need for any cumbersome legacy intermediary. The NFT platform curation process puts this power directly back into the hands of the creator, where they can determine the size of the royalty they wish to receive. 

The Future of AI and NFTs

One impressive thing about the 2021 NFT boom was that it occurred without the use of Generative AI protocols. Digital artists excelled in that environment, but now that anyone can create high-production value art with the ease of a chatbot, it's unclear how lucrative this market will be in the future. More emphasis will likely be placed on the utility and community of projects. 

Generative AI protocols can empower individuals to become great artists themselves and open career doors that had previously been out of reach. However, one of the major concerns that artists ran into during the last cycle was that their art was sold as NFTs without their consent. There remains legal ambiguity over the monetization of digital assets that have been produced through generative AI protocols. These two factors may be on a collision course, especially if Generative AI assets are used to start creating generational wealth through viral NFT collections.

Plagiarism also played an active role in the last cycle when NFTs were duplicated and minted across multiple blockchains. This theme of a lack of interoperability and data siloes has been explored in previous blog posts.This is where AI can play an important role. Through its security enhancements of early anomaly detection and fraud prevention, AI can be a backstop similar to its role in RWAs and DeFi. This is essential in establishing cross-chain interoperable security when determining the provenance of digital assets.