Author: Cookie & alertcat.eth, ChainCatcher
ChatGPT, a chatbot owned by OpenAI, reached 100 million monthly active users just two months after its launch, making it the fastest-growing application in history. Such a powerful ability to "increase fans" has quickly spread the popularity of AI to the encryption field. On January 10, Bloomberg said that Microsoft is considering investing US$10 billion in OpenAI, the developer of ChatGPT. All cryptocurrencies with AI concepts have been completely detonated, including FET and AGIX. Wait for the increase to exceed 200% within a month.
With the help of capital, can these two cutting-edge technologies that have attracted much attention be integrated together? Artificial intelligence uses computers to solve problems by imitating the thinking capabilities of the human brain. OpenAI provides natural language processing (NLP) models with massive amounts of training data to make them more powerful. In the encrypted world built by blockchain technology, the huge amount of on-chain data every day can provide "fuel" for the AI engine, allowing AIGC to feedback better strategies.
Additionally, as AI algorithms become smarter, it becomes increasingly difficult for people to understand how they arrive at decisions and conclusions. The immutable nature of blockchain helps us access an immutable record of the data and processes used by AI in its decision-making process.
Compared with artificial intelligence such as Stability AI and ChatGPT, which have gained a lot of attention and adoption in traditional fields, the greater imagination of blockchain lies in the economic system that can change the AI model. When the FOMO subsides, this article will explore what are the characteristics of crypto projects that introduce AI technology? What kind of chemical reaction can AI combined with blockchain produce?
AI infrastructure
A common feature of AI infrastructure projects is the distribution and sale of traditional AI architecture (data, models and computing power). They generally use their own native token as a medium of exchange. They often occupy an intermediary position between users and service providers, building a decentralized trading market. These are tasks that need to be completed by traditional AI, such as projects in the fields of NLP, AI voice, and CV that use DApp as an intermediary platform for transactions. Essentially, it is a decentralized market that uses token pricing and exchange in the traditional market.
Openfabric AI
Openfabric is a platform for building and connecting AI applications. Through the platform, collaboration between AI innovators, data providers, enterprises and infrastructure providers will facilitate the creation and use of new intelligent algorithms and services. The Openfabric ecosystem consists of 4 roles: algorithm creators, data providers, infrastructure providers, and service consumers. Service consumers need to pay the other three service providers.
Algorithm Creator: Leverage their expertise to create AI algorithms that solve complex business problems.
Data providers: Ensure distribution of large volumes of data required to train AI algorithms.
Infrastructure provider: All the hardware that runs the AI platform.
Service Consumer: An end user who requires a specific business product or service.
Oranges
Oraichain is an AI-powered blockchain oracle and ecosystem. In addition to data oracles, Oraichain aims to become a complete artificial intelligence ecosystem in the blockchain field as a base layer for creating smart contracts and Dapps. With AI as the cornerstone, Oraichain has developed many important innovative products and services, including AI price feed, fully on-chain VRF, Data Hub, AI Marketplace with more than 100 AI APIs, AI-based NFT generation and NFT copyright protection, Royalty Protocol, an AI-powered revenue aggregator platform and Cosmwasm IDE.
Fetch.ai
Fetch.ai is a blockchain platform based on artificial intelligence and machine learning that allows anyone to share or trade data. As an autonomous machine-to-machine ecosystem, any network of independent parties can become a network agent for Fetch.ai, recording any agreements produced between agents on the Fetch.ai blockchain. FET is the native token of the Fetch AI blockchain and is the primary medium of exchange for payment transactions.
Source: Fetch.ai Blog
SingularityNET
SingularityNET is a decentralized artificial intelligence platform and marketplace. Developers publish their services to the SingularityNET network, making them available to anyone with Internet access. Developers can charge for their services using the native AGIX token. Services can provide inference or model training across domains, such as image, video, speech, text, time series, bioartificial intelligence, and network analysis.
SingularityNET Ecosystem
The SingularityNET ecosystem will provide AI services to the platform and create large-scale utilization of the AGIX token. These SingularityNET spin-offs are being developed in multiple strategically selected vertical markets, including DeFi, robotics, biotech and longevity, games and media, arts and entertainment (music), and enterprise-grade AI.
Revisit
The Gensyn Protocol is a Layer1 network for deep learning computation, with instant rewards for supply-side participants who invest their computing time into the network and perform ML (machine learning) tasks. The protocol requires no administrative oversight or enforcement, but instead facilitates task allocation and payments programmatically via smart contracts. The fundamental challenge of this network is validating the completed ML work. This is a problem at the intersection of complexity theory, game theory, cryptography, and optimization. The Gensyn ecosystem consists of 4 roles: submitter, solver, verifier, and reporter.
Submitters: Provide tasks to be computed and pay for completed units of work.
Solvers: Perform model training and generate proofs for verification by verifiers.
Verifiers: Key to linking the non-deterministic training process to the deterministic linear computation, replicating parts of the solver's proof and comparing distances to expected thresholds.
Whistleblowers: Check out the work of validators and pose challenges in hopes of winning the jackpot.
Gensyn's vision is to reduce Dapps' dependence on Web2 infrastructure by providing critical infrastructure components for Web3 applications by decentralizing ML computing.
Application scenarios
In such application scenarios, the project aims to use AI to handle the emerging needs arising from the development of blockchain in recent years.
These needs can be to enable chain game users to skip tedious operations, enable developers to quickly develop chain games, socialize on the blockchain platform, generate virtual people with their own personalities, or detect fake NFT projects, etc. Different from traditional AI platforms, this type of project has strong irreplaceable demand, which gives it a deep moat. At the same time, the difficulty in the development of a platform that uses emerging needs as a selling point is to acquire customers. How to attract enough customers? The need for customers to prove that their platform is sustainable and objective has become a major problem encountered in the development of this type of platform.
Chain travel direction
Under the mainstream financial system of the “P2E” model of encrypted games, users are faced with ever-changing gameplay and a large number of repetitive basic operations. AI can provide players with stable automated processes and develop game strategies with higher winning rates. rct AI is a complete solution that uses AI to provide a complete solution for the gaming industry. Its core technology Chaos Box is an AI engine based on deep reinforcement learning. rct AI has developed an AI training DRL (Deep Reinforcement Learning) model for Axie Infinity. Since the number of combinations of all cards in Axie Infinity is about 10^23, as well as the game's game features, rct AI's model is used in a large number of simulations The efficiency and winning rate have been improved in the battle data.
In addition, AI can provide action prototypes for developers. Mirror World is a game matrix virtual world based on Solana. It has used AI technology to launch Mirrama, a PVP-based arena game Brawl of Mirrors that combines Roguelike gameplay. In addition, Mirror World has also launched a series of NFTs that can be interoperated in the game. The prototypes of these NFTs are completed using AI action algorithms.
Related reading: "Conversation with rct AI: It's time to think about the changes that blockchain will bring to the game publishing side"
social orientation
PLAI Labs, which focuses on leveraging AI and web3 to build a next-generation social platform for users to play, talk, battle, trade and adventure together, received $32 million in funding from a16z in January 2023. Currently, PLAI Labs has demonstrated 2 products to the outside world:
Champions Ascension is a massively multiplayer online role-playing game (MMORPG). Players can choose to own their own characters in the form of NFT, and can fight in large Colosseum arenas, complete quests, and play in custom dungeons. Build and compete and trade digital items.
An AI protocol platform that will help with everything from user-generated content (UGC) to matching to 2D to 3D asset rendering.
PLAI Labs plans to launch the V2 white paper this year, including details of the core economic cycle (using NFT and blockchain to enhance the experience), UGC toolkit (including AI) plans...
Related reading: "Entrepreneurial veterans start again, Plai Labs briefly talks about why they chose Web3"
NFT direction
Aletha AI proposed the concept of iNFT, a technology that combines artificial intelligence and blockchain. After integrating AI, NFT has interactive, generative, scalable and unique personality characteristics.
To put it simply, if NFT is a digital work, after integrating AI, it becomes iNFT, an NFT work with the ability to chat with users. On June 10, 2021, the world’s first iNFT Alice was auctioned at Sotheby’s for $478,800.
Altered State Machine (ASM) is an innovative project that combines NFTs, artificial intelligence, and machine learning to provide training power for AI-driven NFTs. Its vision is to become the ownership and monetization protocol for AI using NFT technology. In the ASM ecosystem, an AI-based Avatar is called an Agent, which consists of a brain and an avatar. The project also issued ASTO tokens to power the ASM ecosystem.
Related reading: "Detailed explanation of Altered State Machine: Innovative exploration of using AI and machine learning to evolve NFT"
Optic is building an AI-powered NFT verification protocol focused on NFT fraud analysis and NFT value discovery within the community, aiming to help the entire NFT market achieve greater authenticity and transparency. The Optic intelligent engine learns real NFT series and then searches NFT collections on the market. Optic then returns a match score indicating how closely the NFT being inspected matches the real NFT.
Optic completed US$11 million in financing in July 2022, led by Pantera Capital and Kleiner Perkins, with participation from Circle Ventures, Polygon Ventures and others. Currently OpenSea has adopted Optic’s Copymint detection service.
Related reading: "A Brief Analysis of Optic: Artificial Intelligence NFT Verification Protocol"
trend analysis
Judging from the current development path of blockchain AI projects, the infrastructure of AI is composed of three parts: data, algorithm and computing power. A normal AI project that wants to realize the ability to generate or analyze artificial intelligence requires a model and data set, as well as a software ontology that calls the model and its GUI. Then there are intermediaries in this field for the distribution of models and data sets, model training (computing power leasing), and software front-end development, and this will give rise to blockchain AI projects designed to efficiently meet customer needs.
For example, as mentioned above, Fetch.ai acts as an intermediary and allows customers to use its native token to trade data sets. SingularityNET allows customers to purchase computing power training services from developers. Openfabric AI customers need to obtain models (algorithms), data sets, infrastructure (software) and other services from providers. Humans.ai is essentially encapsulated in NFT The AI model trained on the data set can be purchased by users using native tokens.
Gensyn is essentially a decentralized computing power rental platform. These are tasks that need to be completed by traditional AI, such as projects that use DApp as an intermediary platform for transactions in the fields of natural language processing, AI speech, and image generation.
Then decentralized applications in the blockchain have created new demands, and AI projects based on the direction of chain games, social directions and NFT are aimed at solving user pain points in the blockchain. For example, rct.ai solves the problem of chain game users Mirror World solves the problem of manual repetitive operations in chain game development, while other projects are developed for blockchain social networking and NFT.
At present, in the initial stage of Web3 social networking, the introduction of AI is more of a narrative method. In the future, there are some possible directions for AI project research and development:
Enhance data privacy: Web3 can maximize data privacy protection by using zk technology, and AI can analyze data without compromising privacy.
Smart contracts: Web3 technology can integrate AI applications into Web3 applications through smart contracts, thereby achieving controllability of the AI model. This type of application can be used in the transaction of models and data sets to realize automated transaction processes and use ZK technology to protect user data. However, this type of project faces the impact of open source data sets and open source models. Just imagine: If users can obtain open source data and models on Hugging Face and use auto train training, why would they trade on the blockchain platform? Facing the impact of Web2 companies, Web3 AI model and data set transactions do not have enough moats.
More efficient machine learning: Web3 technology can improve the efficiency of machine learning in a decentralized manner, making AI applications faster and more reliable. This has been applied in traditional AI training. For example, the improved version of AlphaGo, KataGo, uses distributed training technology, allowing people around the world who want to update this AI to voluntarily provide computing power training. The application in blockchain can be similar For Gitcoin, you can get POAP by donating computing power, or similar to AMM, it provides incentives for liquidity and becomes a platform for renting computing power for a fee. However, due to the high volatility of currency prices, this type of application is less efficient than traditional GPU computing power. Leasing does not have an advantage. Unless the platform itself is engaged in financial business and can subsidize users from the value captured by the agreement, such as Numerai, which uses AI technology to make profits from the stock market, only then will enough users be willing to provide the three elements of AI to enter the platform. .
Summarize
At present, both the blockchain's native AI infrastructure and the encryption projects that use AI engines to implement application scenarios are in their infancy. The main goal is to create a suitable underlying infrastructure and integrate token economics with hardware providers. , data providers, AI algorithms and other artificial intelligence solutions.
Judging from the distribution of the underlying blockchain, AI concept encryption projects are at a “relatively fair” starting line. The encrypted data platform Rootdata currently includes 24 artificial intelligence encryption projects, which are distributed in Ethereum, BNB Chain or its own public chain. No one public chain has a dominant effect. This may be due to the impact of the explosion of the application layer. Leading institutions seem to prefer the exploration of generative AI in NFT and chain games.
However, the integration of the two also faces many challenges. First of all, blockchain trends such as Rollup, ZK and other complex technologies will bring challenges to AI in obtaining data. Secondly, there is not enough continuous experimental data to support the applicability of AI in the blockchain ecosystem and the ability of the AI engine to adjust in response to emergencies. Finally, false projects that use AI concepts frequently appear in the encryption field, making it easy for people to lose confidence in exploring this field.
All blockchain AI projects that solve traditional AI problems need to answer a question: Why does this platform need to introduce tokens on the blockchain? This makes platforms whose trading targets are existing targets in the Web2 market, such as models, data and computing power, suffer from onboarding disadvantages.
Token economics is like a flywheel, which can change the rise and fall cycle of a project. At present, if you want a positive flywheel, you need to consider the actual users of the platform, that is, the problem of customer acquisition. The irreplaceability of demand is the moat of a project. Projects that lack a moat can achieve short-term success, but there will not be enough Users and a strong developer ecosystem. When demand is false, economic incentives are unsustainable and the life cycle of the project will be shortened. We look forward to the emergence of more AI+Web3 projects based on real users and irreplaceable needs. They are designed to fulfill requirements that are not available or poorly fulfilled in web2, so that Web3 needs to be introduced natively.
In any case, the integration of AI into Web3 is a future technology trend, and there are already some examples of Web3 applications that combine artificial intelligence. As time goes by, more related Web3 infrastructure and new models will appear one after another.