AI and Web 3 — A Marriage of Convenience
Breaking news! AI has an attribution problem.
My good friend CHAT GPT helped me describe an attribution problem in regards to the relationship between AI systems and the owners of the information they are trained on:
Why is this important?
With the growing prominence of AI systems in our society, their reliance on a substantial amount of individual and collective data continues to expand. While I believe that the outcomes generated by these AI systems can bring significant benefits to humanity, it is crucial for us to consider the allocation of economic value. Looking at our present world as a reference point, we observe that a large portion of the value created within digital information systems, thanks to the combined contributions of our society, tends to be concentrated among a disproportionately smaller segment of the population. Primarily, it is the platforms and aggregators that display the outputs who capture the majority of this value.
Why is facebook or any other social media giant great? Why do these companies make so much money? The primary reason lies in their vast network of human beings who initiate and sustain conversations and create the content that enables these platforms to flourish. Additionally, as a consequence of these interactions, these platforms gather valuable data about us, which fuels their advertising businesses. However, these companies often receive an excessive share of value compared to the users who are the true driving forces behind these platforms.
This is not necessarily a bad thing; instead, it is something that requires evolution. I am not presenting a moral argument, so I am not asserting that the current state of the value equation is morally wrong or lacks logical justification. Let’s consider what these platforms have achieved: they have created a morally commendable system that eliminates global barriers to information sharing. The creators of these platforms have been duly rewarded for the significance of their work, as, at the time of their invention, they held the highest value.
However, my main focus today revolves around the necessity for a shift in the value equation, now that we have come to recognize the equal significance of the information and content that bring vitality to these platforms. In the context of AI systems, it is essential to share the economic value more equitably between the creators of these AI systems and the owners of the data and content that serve as the training material for these systems.
What does this have to do with web 3, isn’t that a dead thing?
I reached out to my friend Chat GPT again to better explain how web 3 is suited to solving this problem:
Web 3 and blockchain systems, with their cornerstone technology of smart contracts, offer the potential to establish improved ownership systems between platform creators and owners. Smart contracts are self-executing agreements with predefined rules encoded on the blockchain, enabling transparent and automated interactions without the need for intermediaries.
In the context of ownership, smart contracts provide several advantages. First, they enable creators to establish clear terms and conditions for ownership rights. For example, content creators can define specific conditions under which their work can be accessed, shared, or monetized. These conditions can be embedded within the smart contract, ensuring that ownership rights are automatically enforced and respected.
Additionally, smart contracts facilitate programmable value distribution. They allow for the automatic and transparent allocation of royalties, revenue shares, or usage fees to the rightful owners. This eliminates the need for intermediaries to oversee payment processes, reduces administrative overhead, and ensures that owners receive their fair share of the value generated by the platform.
As someone who is not the most technical person in the world, the one thing that frustrates me the most about the web 3 systems that I have interacted with is that they are not very intuitive to the non web 3 native. I am aware that many others have expressed similar concerns, but I firmly believe that in order to fully harness the socioeconomic advantages offered by web 3 systems, they should be integrated as an additional layer to our current web experience. Rather than expecting everyone to comprehend or desire a completely new internet experience, overlaying web 3 functionalities onto our existing framework would facilitate a smoother transition and broader adoption.
Let’s consider OpenAI and Chat GPT as a case study to drive the idea home. Imagine if OpenAI were to implement a smart contract with the owners of its training data. This smart contract would effectively track how the data is utilized and automatically provide rewards to the data owners whenever OpenAI generates revenue from the outputs produced by Chat GPT.
By implementing such a smart contract, OpenAI would establish a transparent and automated mechanism for acknowledging the contributions of data owners. Whenever Chat GPT generates value, whether through subscriptions, licensing, or other revenue-generating channels, the smart contract would trigger the distribution of rewards to the data owners in proportion to their data’s usage or significance.
In our hypothetical scenario, where none of the existing players in the current infrastructure have created web3 on-ramps (assuming this assumption to be true), OpenAI’s initiative would serve as a groundbreaking step towards integrating smart contracts and initiating the transition into the new internet with improved ownership structures. This pioneering effort by OpenAI, in the absence of similar initiatives from other players, would set a precedent for other internet entities to follow suit. It would highlight the importance of incorporating smart contracts to ensure fairer ownership models and empower data and content owners within the emerging web 3 landscape. By taking the lead in implementing such innovative approaches, OpenAI would not only contribute to the broader adoption of web 3 principles but also inspire other entities to explore and adopt similar ownership structures that benefit data and content owners.
I acknowledge that this is a simplistic examination of a complex idea, it is certainly not perfect. Issues about incentives might arise, like are the current platforms who accrue most of value incentivized enough to care about fair distribution of value especially when most of the users are not aware of how valuable their contributions are to these systems.
As my man Charlie M would say:
Apart from incentives, there are other aspects to consider when implementing such a system. It’s not always clear whose data was used to generate a specific output and how much economic value it holds. Additionally, there’s a technical challenge at hand. Let’s be honest, it’s no easy task. However, with some of the brightest minds dedicated to tackling this problem, there’s no doubt that we can find solutions and overcome these hurdles.
