Data Markets Democratize Ideas
Trading and acquiring digital assets goes well beyond NFT shills.
Our relationship with digital assets is a tenous one. One the one hand, NFTs have given awful press to the concept of cryptocurrency and digital asset trading. On the other hand, data assets are often under-valued. It is not always clear how much value a dataset may provide, but all you have to do is look at Amazon, Facebook and Google, to see how profitable consumer data can be. One way to manage these assets, could be to buy and sell them to groups who will best make use of their technology. Going further, opening up this market would enable groups of different cultures and ethnicities to spread their own ideas and values, while being able to gain wealth from it. This is not a marketplace intended for aggressive trading moves, but a platform for society to use to enrich itself.
There are a lot of ways a digital asset could operate. Some typical assets may include:
Access to datasets, given privacy concerns are addressed
Access to an aggregated or synthetic version of a dataset
Licenses to access to Foundational models, the large, powerful forces enabling advanced AI image generation and text generation
Licenses to use high-end GPUs
Prompts, or specific inputs to AI that gives a specific image or text.
Embeddings that give you a way to represent text and images, so that AI can use it solve meaningful problems.
All of these exist only in the Metaverse. In this digital realm, users could purchase and sell embeddings, prompts, and licenses to foundational models, allowing them to use these assets in their own creations and applications. There is also concern over concepts like prompt-trading has, since these prompts mostly consist of a few common nouns and adjectives chucked together- only generating an interesting image due to random chance. This points to the bigger problem: data assets could be distributed to and used by everyone, so why limit access to it? Sadly, it’s inevitable. People have always tried to capitalize on their assets. It took a long time before we realized our data was the source of money for Amazon, after harvesting our data to build up its highly profitable sister company, Amazon Web Services. Already, many artists and content creators are protesting against the use of AI foundational models, citing “art theft”. Others are scraping together money to fund GPUs, so they can leverage the latest AI tech to enhance or complement their works. The idea of this marketplace will put data assets into the hands of the creators who will best use it, and allow creators to generate data assets that the trading community can value democratically.
The ability to trade and monetize these assets would create a market for digital content, allowing individuals and organizations to generate revenue from their creations and innovations. This would encourage users to continue developing and improving their assets, leading to a higher quality and more diverse range of digital content in the Metaverse. For instance, if someone creates a unique and high-quality embedding of a particular ethnic woman, e.g of Central Asian backgrounds, they could sell it to other users who need that specific embedding for their own creations. Individuals of this ethnic background may be granted an automatic stake, so they can gain a share of any revenue generated from its use. This would allow the creator, and the subjects, to generate revenue from their digital asset, incentivizing them to continue developing and improving it. Regulations around these assets could also help with redistribution action, such as offering subjects a share of the proceeds their data helped generate, or set out their preferences for how their data could be used and for what purposes.
In addition to enabling market efficiency, this Metaverse marketplace would also help in data valuation tasks, which is currently an underdeveloped field. The ability to trade and monetize these digital assets would provide a clear and objective measure of their value, allowing individuals and organizations to more accurately assess the worth of their data, and data of potential acquisitons. It would also enable not just possible monetary worth, but information on the demand and usage of the asset too. For example, if one particular foundational model is being licensed out by a handful of startups centered on coffee cultivation, it may that an opportunity for data labelers.
A data asset market, focused on the trading of datasets, embeddings, prompts, and licenses to foundational models in a "Metaverse" would not only enable market efficiency but also help in data valuation tasks by providing a clear and objective measure of their value and demand. This would benefit both creators and buyers of digital assets, allowing them to better understand and assess the worth of their data. It also brings opportunties for social inclusivity, retributive action and support for independent artists, as long as the market is protected from profit-seeking traders with no attachment to the actual digital goods themselves.