When AI Meets Web3

Author: Biyuan Chain

Before 2020, the path for Web3 was paved by creators. However, Web3 encountered finance, which took every creator by surprise; stablecoins, AMMs, and lending suddenly ignited that summer of 2020. After that, Web3 met NFTs, games, and the metaverse. Yet, due to technical limitations, apart from finance, none of these could bring Web3 to the heights once achieved by DeFi (decentralized finance).

Now, Web3 has encountered AI, and the two seem to be a perfect match.

1. AI

AI was once highly anticipated. However, it has always been limited by a lack of sufficient resources for algorithm model development and training, giving the impression of much noise but little action.

But this year, as the singularity approaches, increasingly layered neural networks are beginning to stack, enhancing AI's capabilities, and AIGC is starting to explode with the potential to change the world.

2. Web3

The core of Web3 is blockchain, and the core of blockchain is a decentralized verification network. For a long time, we have emphasized that what blockchain does is regulate production relationships. But this has also been a long-standing issue troubling blockchain: the vast majority of human society's productive forces still originate from the physical world, while blockchain itself is almost oblivious to the physical world. We need various perception devices (oracles) to enable blockchain to interact with external information.

3. AI + Web3

Therefore, the most suitable area for Web3 to thrive is in the productivity generated entirely in the digital space, and AIGC (AI-generated content) comes at the right time.

In the long run, Web3 has very mature verification technologies and rich DAO (decentralized autonomous organization) experience. Through Web3 technology, market principles can be publicly applied to reconcile all relevant stakeholders in the production and consumption of AIGC. "Voting with your feet" embodies market principles and is also the core spirit of Web3. This free atmosphere is also very helpful in stimulating creativity.

Currently, the issue of AIGC infringement is quite tricky. For example, an AI algorithm generated works in the style of Disney illustrator Hollie Mengert using just 32 images, and the model was trained by only one engineer who shared the algorithm model on Reddit.

Infringement must be addressed. The maintenance of copyright is related to the fundamental logic of the entire modern innovation system; condoning infringement essentially suppresses innovation. But the question is: when we combat theft (AIGC infringement), is it necessary to punish the stone used to break the window (the AI model and its operators)? Conversely, how do we incorporate AIGC into the entire copyright protection system?

Ultimately, the core issue is actually just one: the profit distribution problem of AIGC.

Only magic can defeat magic; technical problems (AIGC) can only be solved with technology (Web3). To summarize in a formula:

"Web 3 + AI = Production Relationships + Productive Forces"

● Visionary Network: Creators, sources of creativity, AIGC operators, and blockchain

Through Web3's blockchain technology, it is entirely possible to create an ecological network centered around AIGC. In addition to creators using AIGC, it includes three main aspects of creative tool providers: owners of creative sources (artists), AIGC operators, and blockchain. The fees paid for content generated by AIGC creators can form a DAO for distribution, with the three creative tool parties voting regularly to decide the income distribution ratio.

(AIGC Ecosystem Diagram)

1) AIGC Creators

Creators are in the position of demand and input of value in the entire ecosystem. AIGC helps them complete designs or meet daily work needs more efficiently, and it can even significantly reduce the costs of low-end repetitive designs with AI assistance, achieving a new level of productivity where one person is equivalent to a studio.

2) Owners of Creative Sources (Artists)

They upload their works (such as images, music, videos, etc.) to the Web3 blockchain network for verification and labeling of rights. These works provide material for AI to train new styles. Owners of creative sources provide creativity for the entire system and hold the copyright for uploaded works and styles.

The training process of AIGC is essentially a process of learning from multiple works to form a style, so by reversing this process, we can link the final generated works to the source works. Ultimately, copyright fees will be distributed to different owners of creative sources.

Even when AIGC creators complete their works, they can upload the newly generated works to Web3, becoming new owners of the work styles and enjoying corresponding copyright benefits.

3) Operators: Computing power, storage, rendering, model training, etc.

AIGC operators may include multiple service aspects, such as providing creative computing power, storage space, rendering power, model training, etc. These aspects may initially be provided by a single participant, but as development progresses, different participants may emerge, also participating in the DAO for governance.

4) Blockchain Validators

Providers of verification centered around blockchain offer notarization work for various links in the entire process.

Profit Flow Model

From the perspective of a closed-loop ecological economic model, the income provided by AIGC creators will be distributed among the three parties: owners of creative sources, AIGC operators, and blockchain within the DAO:

"AIGC Creation Usage Fees = Operating Fees + Copyright Usage Fees + On-Chain Verification Fees"

In Summary

Blockchain Web3 can blend the four parties (creators, sources of creativity, AIGC operators, and blockchain) in the form of a DAO, achieving a mutual match of production relationships and productive forces.

In this light, AI and Web3 seem to be a match made in heaven.

How does Maze integrate with AI?

Maze's vision: Enable every user to create, use, and trade NFTs within Maze.

To achieve this vision, through blockchain technology, the use and trading of NFTs can already be accomplished well, but there are certain bottlenecks in the creation and generation of NFTs. Artists face certain entry barriers, and ordinary users cannot create excellent NFT works.

Maze AI can effectively solve the creation problem of NFTs, allowing users to reference artists' works or directly generate excellent works using descriptive terms, thereby greatly lowering the barriers to NFT creation and fostering a flourishing landscape.

By generating a Maze NFT material repository from artists' works, users can select and reference artists' styles to generate NFTs. When NFTs are traded, royalties can be shared with artists, thereby better incentivizing their participation.

The entire process is illustrated in the diagram below:

Maze NFT: The Maze NFT platform for generating, storing, circulating, and trading NFTs. Initially, the official team will collect a repository storing a large number of NFT generation materials, which will be used to create new NFTs and train AI models. Later, artists and third-party operators can also provide corresponding materials and services, enhancing functionality and expansion.

Maze AI: The core of the entire system, primarily integrating AI model training, image generation, NFT construction, and other processes. The services can be provided by the official team or integrated with other third parties, allowing users to conveniently generate resources and NFTs using this AI service.

Artists: Artists can upload their works to the NFT repository as materials for AI generation. If users generate NFTs in the style of that artist and trade them, the artist can receive corresponding shares.

Ordinary Users: Users can select their favorite materials from the gallery, use Maze AI to imitate and generate new NFTs, and list them on Maze for trading and circulation.

Planning

Phase 1: Establishing Maze AI and Generating Content

The official team has collected a classic style library to add to the NFT repository, and through the galleries of resident artists, a basic library has been formed for users. Maze has streamlined the entire process from gallery, AI generation, NFT minting, to NFT listing. Users holding NFTs like the Three Kingdoms and $BTM can gain more rights as early users.

Phase 2: Incentivizing the Closed Loop and Prosperous Ecology

Setting copyright fees, usage fees, gas fees, and other revenue dividends, distributed through smart contracts to ensure transparency. This greatly incentivizes the participation of artists, computing power, rendering, and storage operators, providing richer artistic styles and creations, stronger rendering and storage capabilities, and more diverse AI training models. This allows users to experience richer and more powerful functionalities, creating a positive cycle that greatly enhances the ecology.

Phase 3: Exploring More Possibilities of AI and Web3

Beyond AIGC, Web3 can have other points of integration with AI. For example, the combination of AI models and blockchain identity verification (avatars, PFPs, domain names, etc.), as well as using blockchain's decentralized network to transmit core information and then using AI for local generation, or using AI to assist blockchain verification—these are all long-term exploration directions.

Maze has never stopped exploring.