Yesterday, when I reflected on the main narrative of this round of market in the article, I couldn't help but mention AI again. When it comes to AI, I suddenly thought of a new model that emerged this year to use crypto assets to finance AI: IMO (Initial Model Offering).
It was proposed by ORA (ora.io), an AI project based on Ethereum.
Before explaining IMO in detail, let's review the core elements of AI.
There are three core elements of AI: computing power, models and data.
These three elements are currently all monopolized by centralized institutions/companies:
Nvidia has a monopoly on computing power.
Internet giants led by OpenAI have a monopoly on models.
Search engine giants have a monopoly on data.
This monopoly situation has always been something that entrepreneurs in the technology industry, especially in the crypto ecosystem, do not want to see. Some of the so-called "AI + Crypto" projects that have emerged in recent years are committed to solving the monopoly problem of these elements and hope to decentralize them.
However, each of these elements requires huge capital investment. Even if decentralization is finally achieved, a lot of capital is still needed in the start-up phase of the project.
IMO aims at providing funding for the project at the start-up phase.
But it only focuses on solving one of the three elements: the model.
So what is IMO?
According to the project party's explanation, it is a means of financing AI models by tokenization.
Speaking of tokenized financing, investors who have experienced the bull market frenzy in 2017 will immediately think of 1CO.
That is a means of financing projects with Ethereum ERC-20 tokens:
The project party sells the project tokens at a certain price, and investors buy these tokens with Ethereum. After the project party gets these Ethereums, they will have the capital to develop and promote the project. Then the token holders can share the rights and interests represented by the project tokens.
IMO is not essentially different from 1CO in form, but the financing objects are different: 1CO financing can be used to develop any crypto project, while IMO financing is specifically used to develop AI models.
According to ORA's detailed explanation of IMO, it tokenizes open source AI models to provide financial support for the continued development of the project and achieve community-driven development. The project party implements IMO on the AI model and issues the corresponding ERC-20 tokens. Investors participate in the purchase of these tokens and enjoy the rights and interests represented by the tokens.
This application has no major innovation in form, but it still has certain characteristics in terms of scenarios. It at least provides a new financing idea for the development of open source AI models.
I don’t seem to have seen any AI model team use this method to raise funds before.
But there are still many questions to be verified whether this idea is feasible:
- Is the capital raised in this way enough to start the development of an AI model?
- Even if there is a team that develops AI models in this way, where is its use scenario? In the case that Apple has begun to integrate OpenAI models, will this type of model have an advantage over OpenAI models?
- Existing AI models have not generated visible commercial value so far. Can this model generate commercial value?
- Even if this method can succeed, it will sooner or later hit the regulatory iron plate once it is rolled out and expanded. What should be done then? Unless the model it creates can quickly grow to a position as difficult to shake as Ethereum.
The above questions still assume that this model is used in positive situations. If this model is also used in a large number of mixed situations, the consequences may be as disastrous as 1CO.
In short, it still has too many uncertainties and high risks.
However, any innovative idea is always full of doubts and challenges when it comes out. Although this project has a lot of uncertainties, in my opinion, it is at least better than a lot of so-called "AI + Crypto" that is dressed in AI but actually has nothing new.