When Web3 encounters AI
2022-11-11 15:00
Maze NFT 交易平台
2022-11-11 15:00
订阅此专栏
收藏此文章

Web3’s road was paved by its founders before 2020. However in that Summer, Web3 encoutered with finance, which exceeded all founders’ expectation. The following stablecoin, AMM and lending protocol lighted that summer’s night like a firework. Later on, NFT, game and metaverse all had sparks with Web3, but due to technical limits, none of them expanded Web3 like Defi (dencentralized finance) had done.

However, Web3 encounters with AI this year fortunately, which seems like a heaven-made match.

1. AI

AI was highly placed great hope as well. But its ability has been controversial for years,limited by lack of sufficient resources for algorithm model development and training.It always impressed us with thunder but not rain.

This year the singularity is at the corner. More and more layers of neural network are integrating, and AI begins to outbreak the potential of changing the world,so does the AIGC (AI Generated Content).

2. Web3

Web3 is founded on blockchain, which is basically a decentralized proving network. Despite its mission to establish new production relations, blockchain has its own flaw, which is the isolation from physicial world. Traditionally, to link up on-chain and off-chain worlds, sensing devices like orcale are essential to blockchains.

3. AI + Web3

Therefore, the home court of Web3 is still the digital world, while AIGC happens to be 100% digital contents.

Considering the more promising picture, Web3 has very mature proving systems and DAO (Decentralized Autonomous Organization) experience, which could use market principle to balance all AIGC-related paticipants’ interests. Participants could “vote with your foot”, which is not only the market principle, but also one of the very key laws of Web3. With that freedom, the combination of Web3 and AIGC could inspire creatibity to a new level.

As for now, the biggest problem of AIGC is copyright infringement. A few days ago, an AI-model engineer trained his model with 32 artworks of Hollie Mengert, a Disney illustrator, and successfully AI-generated pictures of that style. Though the engineer made no profit so far and just uploaded his model to Reddit, his behavior is hardly unquestionable.

Copyright infringement is absolutly a crime. The defense of copyright is the foundation of modern innovation system. In other words, any indulgence of copyright infringement is the enemy of innovation.

But the question is: when we are against killers, is it right to blame the blade as well(AI Models and Operators)? Or how could we take AIGC into the modern innovation system?

It all leads to one question: how to allocate AIGC’s benefits.

Magic could only be defeated by magic. The problem technology (AIGC) brought could only be solved by technology (Web3) as well.

To sum up with a formula:

“Web 3 + AI = Prodcution Relation + Productivity ”

  • A network of creators, original artwork owner, AIGC operator and blockchain

With the blockchain in Web3, it is possible to create an ecosystem around AIGC. The Creators, Original Artwork owners, AIGC Operators and Blockchain Validators are totally accessible to it. The benefits generated by creators could be allocated to the other 3 kinds of participants in a DAO, where shares are determined by regular voting power.

(AIGC Ecosystem)

1) AIGC Creators

AIGC Creators use AI to generate contents and input value to the whole system. And they also benefit from the system by using AIGC to assist their daily or special design work. Top creators could be saved from repetitive work and focus on their adept parts, even the level of “one designer, one studio” is at their reach.

2) Original Artwork Owners (Artists)

Artists provide ideas for the system, by uploading their artworks (paintings, musics, videos…) to Web3 where these artworks are verified and become NFTs. These NFTs would be used to train new styles in AI models. In return, artists have the copyrights and benefit from it.

The machanism of linking AIGC to orginal artwork is inverting the process of generating. To be more detailed, the generation process of AI model is that AI “merges” different artworks and generates a new one. So theoretically, any AIGC could be traced to its orginal artworks. By that means, the royalty AIGC’s creators’ payments could be allocated to all origianal artworks’ owners.

Original artworks could also come from AIGC. After a new AIGC is uploaded to Web3, it could be the new original artwork and have its copyright.

3) Operators: computing power, storage, rendering, model-training,etc

The operation of AIGC involves many aspects, including computing power, storage, rendering, model-training, etc. Those aspects might be integrated in one platform at first, but these tasks could also be undertaken by seperate participants later on. Likewise, those participants could join the DAO and govern with votes and proposals.

4) Blockchain Validator

Blockchain Validators would satisfy the needs of verfication for the whole system.

Value Flow Pattern in the System

As for the business model, value is inputted from creators, and allocated to original artwork owners, AIGC Operators and blockchain.

“Creation Fee= Operation Fee + Royalty + Gas Fee”

In a nutshell

All participants (AIGC Creators, Original Artwork Owners, AIGC Operators and Blockchain Validator) could be integrated in a DAO through Web3, in order to balance productivity and the relaitons of production.

In that case, Web3 is truly AI’s match.

How does Maze combine with AI?

Maze’s Vision: Create, Use and Exchange NFTs in Maze World.

To achieve this, Maze uses blockchain technology to provide 3 NFT functions: creating, using and exchanging. The latter two functions have been already well implemented. But there is still a bottleneck: not all the users can create wonderful NFT images.

Maze AI appears to be the last piece of the puzzle. By refering to artists’ style or directly inputting prompts to generate NFTs, users could start their creation right away.

If a user chooses an artist’s style in Maze NFT’s material store to generate NFT, royalties will be awarded to that artist when the NFT is exchanged.

The entire process is shown in the figure:

Maze NFT: the platform to generate, store, transfer and trade NFTs. At early stage, Maze AI official will set a material store to store massive NFT materials for new NFTs’ generation and AI-model training. Later, artists and operators can provide their own materials and services to join the ecosystem.

Maze AI: the core of the entire AIGC system including AI model training, image generation, NFT minting… These services can be provided by Maze AI official integally or with other operators jointly, depending on which mode is available and benefits users more.

Artists: Artists can upload his/her artworks to material store and form a style. When a user adopts that style and makes profit from it later, the artist can get his/her corresponding royalty.

Users: Users can generate NFTs from material store and list their NFTs in Maze.

Roadmap:

Phase 1: Build Maze AI

Maze has built the basic material store from two major sources:
1) Classic styles
2) Contracted artists’s works and styles.
And then develop the procedure from store, AI generation, NFT minting, to NFT listing. Those who hold CryptoSanguo NFT and $BTM will obtain corresponding benefits as early supporters.

Phase 2: Enrich the ecosystem

Maze would allocate royalty, operation fee, gas fee… through smart contracts to ensure transparency. It will inspire more artists, computing power, rendering and storage operators to be part of the ecosystem. More artist styles and works, more powerful rendering, larger storage space, and superior AI training models would achieve a mutual beneficial situation where 1) users can experience more functions, 2) all participants can have a fair share on user’s creations, 3) the ecosystem is open and prosperous.

Phase 3: Further Exploration of AI + Web3

Except for AIGC, Maze would explore futher frontier of Web3 + AI, such as the combination of AI model and blockchain identity verification (avatar, PFP, domain name, etc.), local AI-generation from seed informaiton through blockchain network, AI’s assistance in blockchain’s verification…

There is always a next stop for Maze.

【免责声明】市场有风险,投资需谨慎。本文不构成投资建议,用户应考虑本文中的任何意见、观点或结论是否符合其特定状况。据此投资,责任自负。

相关Wiki
Maze NFT 交易平台
数据请求中
查看更多

推荐专栏

数据请求中
在 App 打开