Original author: Riyue Xiaochu
The AI sector is different from all other sectors. Because AI is not only the AI of the cryptocurrency world, but also the AI of all mankind. AI is changing human society vigorously. Maybe in a few decades, our descendants will call it the fifth industrial revolution of artificial intelligence.
For a long time to come, AI will be the biggest focus of mankind and will also be a hot topic in the cryptocurrency circle.
Since the advent of GPT 3.5, the capabilities of AI have shocked the world and will also affect the cryptocurrency industry. The release of GPT 3.5 in 2023 and Sora in 2024 both triggered hype in the AI sector of the cryptocurrency industry. Now, the giants are launching a technology competition to compete for the right to speak in the future of AI. All fields of human society are using AI to innovate industries.
Therefore, there will definitely be projects worth tens of billions of dollars in the AI track. Of course, there are also a large number of projects that want to take advantage of the hot spots to reap a wave of profits. Therefore, we need to be discerning and select truly valuable blockchain AI projects. The following are the important directions I think:
1. Decentralized AI Network
The significance of decentralized AI networks is 1) to prevent centralized giants from having too much power 2) to also be a collective power, bringing together excellent small and medium-sized teams.
Bittensor is a decentralized artificial intelligence network. Anyone can produce competitive digital goods on it. These digital goods can be machine intelligence, storage space, computing power, protein folding, financial market predictions, etc., and receive TAO rewards.
The architecture of Bittensor can be divided into two layers. The bottom layer is a blockchain based on Polkadot Substrate, which is responsible for executing the consensus mechanism and incentivizing the network. The AI layer is responsible for reasoning, training, and ensuring input/output compatibility between Bittensor protocol nodes.
The core of Bittensor is subnets. Subnets are a competitive market. Anyone can create a subnet, or participate in an existing subnet. You create a subnet by paying a registration fee (in TAO). Subnets will compete with each other to achieve survival of the fittest. Current subnets include large language models, fine-tuned models, language translation, data, etc.
For details about the subnet, please refer to: https://foresightnews.pro/article/detail/57931
2. Decentralized computing power
In the AI race, decentralized computing power must have a place, and a very important one. AI has three major elements, data, algorithm model and computing power, and computing power is the only one that belongs to hardware. General hardware needs to follow Moores Law, that is, the performance must double every 18 months to meet market demand. However, we are now in the era of AI for all mankind. The AI competition among technology giants and the implementation of applications have led to a huge demand for GPUs, and the supply of computing power is obviously unable to meet the explosive growth in AI demand. The core value of the decentralized computing power platform lies in using blockchain technology to gather idle AI computing power around the world to meet the markets demand for GPUs.
At the same time, the chicken-and-egg problem can also be solved in the cryptocurrency world. By taking advantage of the expectation of token airdrops, a large amount of computing power can be attracted in the early stages of the project, and after the project is launched, the tokens produced can be used as incentives. We are accustomed to this gameplay, but the reality is that this gameplay can only be done in the cryptocurrency world.
Another advantage is that compared to the GPU service of the cloud center, the use of the decentralized computing power platform does not require very complicated contract process review, etc. There is also an advantage in the price of use, which is only about 1/5 to 1/10.
Akash and Render are veterans in this field, and they have their own advantages. However, in terms of the number of GPUs connected, the two new players Aethir and io.net have far surpassed them. According to data from layergg, Aethir has more than 40,000 GPUs, and io.net has more than 22,000. Akash and Render only have hundreds or thousands of GPUs.
In terms of market value, io only has 260 million USD and FDV is 2.2 billion USD. ATH has 320 million USD and FDV3 is 500 million USD. However, unlike other high FDV, a large proportion of io and net are mining outputs, which are used for mining rewards for decades. Excluding this part, ios FDV is 1.3 billion USD and ATHs FDV is 1.7 billion USD.
3. Web3 Platform for AI
In addition to the general language models of giants, there are still a lot of practical needs that are not met, such as certain data-sensitive industries (such as medical and education industries, etc.), the market experience accumulated by companies, the professional knowledge possessed by teachers or experts, and algorithms optimized to solve specific problems.
There are two basic demands for these needs: one is to protect data or knowledge, and the other is to get economic returns. The blockchain-based AI platform can perfectly meet these two needs.
KIP Protocol is a secure and efficient decentralized Web3 underlying protocol that provides AI creators (data owners, algorithm model makers, AI application developers) with
1) Guaranteed ownership of data/knowledge. Stored on-chain in the form of ERC-3525 semi-fungible tokens to represent ownership of knowledge assets and token gate access rights. They are uniquely controlled by the owner.
2) Data/knowledge usage protection. KIP Protocol is the worlds first Web3 protocol that supports decentralized RAG (retrieval-augmented generation) solutions. RAG is a complex and innovative technology used in generative AI that enables AI models to generate answers that they did not know by retrieving data and information from external knowledge bases and databases.
3) Application layer of AI. Creators can easily access the path of their works through KPI, as well as build front-end code and promote their products. Anyone can easily use various AI applications on the KIP platform. At the same time, a set of transparent and anonymous web3 contracts are used to record the usage of various applications, which is open and transparent.
4) Settlement layer. Kip uses account abstraction to allow users to participate in and use multiple applications without signing multiple gas. Tokenization in the form of ERC-3525 SFT is then used to record and settle transactions between users and token owners associated with each element they interact with. Kip platform tokens can be used for borderless settlement.
On the 17th, KIP Protocol announced a milestone partnership with Moemate. Moemate uses the KIP protocol to build decentralized infrastructure. This partnership will bring more than 3 million users to Web3.
Moemate is a leading AI agent platform that allows users to create custom AI companions with adjustable personalities, multi-language support, and features such as voice cloning and image generation, such as chatbots, discord/telegram bots, AI NPCs for gaming, virtual companions, and even AR/VR fields.
4. Real Human Protection
With the development of artificial intelligence, AI is no different from real people. This is not only reflected in language expression, but also in pictures (including photos of people), videos, etc. On the Internet, you can no longer distinguish whether the account is a real person or an artificial intelligence. Therefore, a serious problem may arise in the future: how to prove that you are a real person on the Internet? At the same time, we also need to distinguish between real people and AI on the Internet, and monitor the potential adverse effects of AI in a timely manner. Then blockchain technology can be used as a proof of real person identity. It can encrypt and store biometric information and ensure that it cannot be tampered with. It can also be used to prove the real person without leaking details.
Regarding this issue, Sam Altman, the founder of OpenAI, who first proposed it, and founded Worldcoin, which aims to establish a global digital identity and financial network owned by humans. Global registration of each individual is achieved through Orb biometric scanning to reliably distinguish between humans and machines. Worldcoin creates Layer 2 with Op stack, and all human biometric information will use Zero Knowledge Proof (ZKP technology) to ensure privacy, and no third party will directly obtain it. The incentive mechanism of the protocol uses a dedicated blockchain token called WLD. However, from the secondary point of view, the circulation of WLD is 800 million US dollars, and the FDV is as high as 28 billion US dollars. Therefore, it is not suitable for long-term holdings, and only consider holding when the AI sector is hyped again.
Reference articles: