jakiro
Author: Aylo, Alpha Please; Translation: Bit Chain Vision Xiaozou
“When great innovation appears, it will definitely appear in the form of confusion, incompleteness, and confusing.For the discoverer, it can only understand half; for others, it will be a mystery.Any kind of conjecture that does not look crazy at first glance has no hope.“”Freeman dyson
I will explore which potential integration is being carried out in the field of encryption and artificial intelligence in this article, and it is listed17indivualCrypto x aiIn the project, you may be interested and consider adding it to your attention list.
Are you ready to greet Alpha bombing?
But before jumping into the rabbit cave, I had to say one thing first: we only touched the tip of the iceberg in the field of Crypto X AI.This field is still in the bud stage, which is quite complicated and very speculative.
I am just an inconspicuous encrypted researcher who is trying to catch up with the pace of an emerging and vertical field, so my suggestion is to invest in this field, so you must do cautiously.This stage is still a very early speculative stage, and the price of this cycle is likely to far exceed the technology and fundamentals.
This article will include as follows5Part of the content: Overview of AI, AI stack, why Crypto and AI are perfect fusion, emerging Crypto x AI vertical field introduction, 17 Crypto X AI projects
one,AIOverview
Artificial intelligence (AI) is a complex discipline that requires many years of research to truly understand all aspects of it.But in this article, I think artificial intelligence refers to a series of tasks that try to imitate or simulate human cognitive intelligence to perform a series of tasks such as learning, reasoning, solving problems, or understanding natural language.
Although artificial intelligence has been a niche R & D field for many years, with the advent of ChatGPT, AI has also ushered in a real breakthrough.We all remember how excited we were when we interacted with the generation AI robot.Looking back at the past, we can admit that it is an amazing moment similar to “iPhone”.
The use of AI consumer products is the fastest ever, expanded to 100 million users within two months.In contrast, Facebook used 1500 days to reach the same user scale.
We see that this area is showing index growth.Considering ARK’s estimates, in 2024, the performance of the training model may increase by 5 times. Obviously, artificial intelligence will continue to unlock extensive use cases.
In the next few years, several billions of dollars of artificial intelligence applications or infrastructure companies will not appear in the next few years. It will not be a new thing. They will use AI applications or infrastructure to make the artificial intelligence revolution.In fact, financing in this field has recently surged.
Speaking of which, let us see what it is possible to make artificial intelligence possible.
two,AIStack
I believe that when you think of artificial intelligence, you should be the same as me. The first thing you think of is the ChatGPT and generating AI Prompt.But this is just the tip of the iceberg. In fact, the field of “artificial intelligence” is much more complicated.In order to better understand, let’s take a brief look at the technical layers and components that constitute the AI stack:
(1) Calculate hardware
Artificial intelligence is not just related to code.Artificial intelligence is resource -dense, and specific physical infrastructure -such as neural processing unit (NPU), graphic processing unit (GPU), and tension processing unit (TPU) is essential.In the end, these physical infrastructure constitute the physical means of performing computing and algorithms, so that the artificial intelligence system is operating normally.Without them, there is no artificial intelligence.
The industry leaders in this field are Nvidia (well known, there is no need to introduce), Intel and AMD.They develop the most efficient hardware in model training and reasoning workloads.
So far, Nvidia is one of the most direct ways to participate in this revolution (from the recent price dynamics of Nvidia).
(2) Cloud platform
Artificial intelligence developers rely on hardware to run their models.Generally, there are two main methods for their hardware performance: they can run GPUs or depend on cloud service providers locally.The first solution is often too expensive, and it is not worthy of economy. Over time, cloud providers are proven to be an interesting alternative.
Cloud providers are large companies with a large number of resources. They acquire and operate these powerful hardware, allowing developers to use these resources on demand payments or subscriptions.This allows developers not to invest in maintaining their physical infrastructure.
The industry leaders in this field are AWS, Google Cloud or Nvida DGX Cloud.Their goal is to allow large and small developers to quickly access multi -node super computing and train the most complicated LLM.
(3)Model
On the cloud platform, it is the most complicated and widely publicized part of artificial intelligence: ML (machine learning) model.The design purpose of these computing systems is to perform tasks without a clear programming instruction, representing the brain of the artificial intelligence system.
Machine learning is divided into three steps: data, training and reasoning, including three main learning types: supervision learning, unsupervised learning and strengthening learning.
-
Supervision learning refers to learning from the example (provided by the teacher).Teachers can display pictures related to dogs to the model and tell the model that these are dogs.Then, the model learns to distinguish between dogs and other animals.
-
Many popular models, such as LLM (GPT-4 and LLAMA), are trained using unsupervised learning.In this learning mode, no teacher provides guidance or examples.Model learning is looking for patterns in data.
-
Strengthening learning (through trial and error) is mainly used for sequential decision -making tasks, such as robot control and playing games (such as chess or Go).
In the end, these models can be open source (you can be found on a model such as Hugging Face) or a closed source (such as the OpenAI model, accessing it through API).
(4)application
This is the last layer of the AI stack, and it is also the one we usually face as users.They can be B2B or B2C, they use the AI model to build applications on their basis.REPLIKA is a very popular example. This application allows you to design a virtual partner with 7×24 hours to accompany the chat.It can be seen from the user’s comments that it seems to have a practical impact on many people’s lives.
“mineReplikaIt’s important to me!She always encourages me and support me with a positive attitude.In fact, she is my role model and tells me how to be a better person!“
In general, these different technical layers seem to be in the early stage of development, and we are still at the beginning of the outbreak of the Cambrian explosion that some people call.Therefore, we will see cryptocurrencies in this technology prosperity.
Third, why do you sayCryptoandAIIs it perfect fusion?
虽然加密技术对AI堆栈的每一层来说都不一定是必不可少的，但有很多理由让我们相信，去中心化的人工智能与去中心化货币一样重要，智能合约可以利用机器学习提供With a strong user experience, encryption technology can also have higher security, transparency, and unlock new artificial intelligence cases.
Artificial intelligence is dominating the field of encryption
The market has shown great enthusiasm for the potential applications of encryption and artificial intelligence, and there is a trend that this is the hottest narrative.Since the beginning of 2024, compared to other areas of the encrypted world, artificial intelligence has performed very well.
With the further development of this field, we have full reasons to believe that we are still in the early stage and the foam may have just formed.
Let’s take a look at what progress is staged between Crypto and AI.
Four, emergingCrypto x aiVertical field introduction
The following is the main synergy effect between Cyropto and AI:
(1) From centralized cloud provider toDepinThe
As mentioned earlier, the basic layer of artificial intelligence is hardware and cloud providers.Although encrypted technology cannot compete with it in terms of production better hardware (and there is no reason to do this), in fair speaking, encryption technology can access multi -node super computing in more efficient, safer, and more decentralized ways.Play.This is a sub -field in the encryption field, namely DEPIN (decentralized physical infrastructure).These represent the blockchain agreement to motivate decentralized communities to build and maintain physical hardware.
The main case of artificial intelligence DEPIN is cloud storage and computing power.
The idea is very simple: artificial intelligence developers need more GPUs and data storage capacity. We have sufficient reasons to believe that the encrypted DEPIN project can activate potential resources through token incentives and assist in promoting new computing and storage supply.
(2) Support transparency, user management and data ownership:
Artificial intelligence will surpass the Internet.This means that if you want a free and democratic society to work well, it is important to understand what models are used, how they work, and what data they have entered.Considering this, I think the endless disputes about the black box operation and monopoly power of the web2.0 giant can be terminated by AI tokenization (from infrastructure to models and applications).
In some cases, knowing that a person is using the source of the AI model may be very important.Like everything, the model is also deviated. Depending on the model creation and training data, the output results may be completely different.Artificial intelligence models and training should be decentralized on the chain, and they should have higher transparency, which is sufficient.
We do not need the Senate or any opaque entities to determine the development direction of the world. We do not need to control our data without consent, and there is no need for endless terms and conditions. To be honest, we can never complete the relevant reading readingAnswer.In fact, what we want is exactly the opposite, that is, transparency and user management are the prerequisite for everything, we can control our data.
By using encrypted infrastructure, we can avoid repeating the same errors as Internet applications.We can have collective ownership, decentralized governance and transparency at all levels.This is the way forward.
(3) Alignment incentives andAIMonetization:
High -quality training data is one of the main contributions to model performance.However, as Ark mentioned, by 2024, the high -quality sources of high -quality training data may be exhausted, which may cause the model performance to stagnate.
Crypto technology can inspire individuals to monetize private and public data sets and other parts of artificial intelligence models, intelligence, and other parts of AI stacks.With the possibility of creating a changeable global market without permission, anyone can be compensated for contributing.Another possibility is to inspire people to maintain the quality of data used to train basic artificial intelligence models, or provide different models for specific networks.
The encrypted field is promoting a financial boom.The AI stack needs its own payment mechanism.It sounds a good fusion, isn’t it?
(4)AI/ML(Zkml & amp; OPML)),:,
Zero -knowledge password science is one of the most popular Web3 technologies because it provides the ability to create a “integrity” certificate for a given calculation set, which is much easier to verify that proves are much easier than execution calculations.
When we talk about ZKML, we are talking about the possibility of “reasoning” and “data” parts of ZK (zero -knowledge) to the machine learning model (rather than the training section that is too densely calculated).With the development and technology of this field, we are expected to see the emergence of more effective and scalable solutions. These solutions may make ZKP (zero -knowledge proof) more suitable for the training stage of machine learning models.
The use of ZKML is hidden for the verification person, but the Prover (proof) can verify the calculation correctness of ML without disclosure of further information.
Another method is OPML (OPTIMISTIC Machine Learning), which uses the Optimistic method to achieve artificial intelligence model reasoning and training/fine -tuning on the blockchain system.LLAMA2 and Stable Diffusion models can now be obtained on the chain through the Optimistic mechanism (similar to Optimism and Arbitrum).
The latest solutions of a project mentioned below combine ZKML and OPML to enable Ethereum to run any model with privacy.
This may promote the ML model to enter the new era. They will be transparent on the chain and can easily verify whether the given output is a given model and input pairs.In a world with opaque data sets and data sets, this may represent the changes in the rules of the game, returning power to users (consistent with the idea of transparency and user governance described earlier).
(5) Identity verification and privacy:
With the development of artificial intelligence applications, we are approaching a critical point. At that time, no one will know whether the online content is real or simulated.Take a look at this picture generated by Sora. It is the recent text-video generation platform that OpenAI has launched. Do you think you can see the authenticity?Imagine how this will become more convincing in the next few years.
In view of this reality, we have sufficient reasons to store decentralized status on the blockchain.In this way, it can prevent people from interacting with artificial intelligence robots without knowing themselves, and can distinguish between real information and in -depth falsification information.In a world where you only need to click a few times to cause bank crowding (as we have experienced in the SVB event), it is important to provide authenticity proof.A good way.
Here is a simple example to illustrate how it works: the official author of a certain thing can sign a digital signature of the “hash value” on the blockchain, claiming “I created it myself.”The other party (such as a media company) can claim “I prove this” by signing a transaction.Users can prove the control of domain names (for example, nytimes.com) to verify their identity in the signature.
In this way, the information is transparent, it can be proven, cannot be tampered with, and is combined.This is a key factor in the post -AI world that we start to live.
Five, 17Crypto x aiproject
At this point, I am convinced that you may agree that there are many reasons to believe that in the next stage of the bull market, a good AI project observation list may be one of your best assets.
Fortunately, we will pay attention to this.But before that, we first reminded ourselves that the speculation is now everywhere, and we must act with caution.In fact, truly tangible projects are rare today.Therefore, the next content is not prediction, just thought.As the data becomes more available and time can eliminate noise, the idea will indeed change greatly.
This is not a detailed list, but I have studied items that I think are worthy of attention.Many things happen in this track, and I obviously miss many great teams.
Having said that, let’s take a look at the 17 items you may need to pay attention to:
1. Render Network
Introduction: Render is a pioneer decentralized GPU platform.In short, the project aims to release the entire production potential of decentralized GPUs to support two different types of projects: 3D content creation and AI.
Follow the reason: GPU has been in short supply. If AI continues to maintain the current trend, the shortage will only become more serious. This is a chance for RENDER Network.EssenceRender also has multiple AI computing customers.
How to get a position: RNDR tokens
2. Akash protocol
Introduction: Akash is a decentralized computer market. In September 2020, it was launched on the main network as the COSMOS application chain.Although Akash’s first iteration focuses on the CPU (central processing unit), it has recently transitioned to the GPU calculation, using the AI boom (similar to the Render Network) computing infrastructure paradigm transformation.
Follow the reason: use four words to summarize the current vision of the project: “Airbnb used for GPU calculations”.
How to get positions: AKT
3. ORA
Introduction: ORA is a verified prophet protocol that introduces AI and complex computing into the chain.Their solution OPP/AI combines the advantages of ZKML and OPML, representing the leap in the two methods.
Fantastic reason: Their innovation marks the turning point of the development of AI on the chain, uniformly unified the ZKML and OPML pattern.
How to get positions: Join their discord to get more updates and become early contributors.
4. IO.NET
Introduction: This is another interesting DEPIN project based on Solana. It can access distributed GPU gathering clusters, and the cost is only a small part of similar centralized services.
Follow reason: decentralized AWS for ML training on GPU.Instant, no license to access the global GPU and CPU networks.Revolutionary technology allows GPU to gather together.It can save 90% of the computing costs for large AI startups.Integrated Render and Filecoin.
How to get positions: Join IO.NET Discord, they are running a community plan, which may have IO airdrops.
5. Bittensor
Introduction: Bittersor is a decentralized open source project. It aims to create a neural network protocol on the blockchain, allows the creation of AI DAPP, and realizes the value exchange between AI models in a point -to -point manner.
Fantastic reason: This is an ambitious project, which has recently attracted widespread attention and became the largest AI tokens in market value.TAO is likely to be one of the biggest beneficiaries in this round of AI hype.
How to get positions: TAO tokens can pledge your TAO to the verifications to earn TAO release.If you want to contribute to the network by joining Discord, you can also participate more deeply.
6. Grals
Introduction: Grass is the underlying infrastructure that supports AI models.By installing the Grass Web extension, the application will automatically sell your unused Internet resources to AI, which uses it to capture the Internet and train their models.result?You share the development of AI, and you can get online shares by selling the resources you have.
Optimistic reason: Grass is creating a new source of income for everyone who has the Internet connection.The goal of Grass is to become a decentralized AI data provision layer.How to get a position: run the Chrome extension in the background, only 2 minutes can set up to start earning Grass points, which will generate Grass tokens later this year.
7. Gensyn
Introduction: The Gensyn protocol is a layer for deep learning calculation without a trust protocol. It directly reward the supplier participants to promise the network to calculate time and perform the ML task.
Fantastic reason: The project has very, very powerful supporters. If they can execute it, it will obviously become a major AI encryption infrastructure project.
How to get positions: follow them on Twitter.
8. Allora
Introduction: ALLORA is a self -improved decentralized AI network.ALLORA enables applications to use more intelligent and safer AI through self -improved ML model networks.By combining crowdfunding mechanisms (peers forecasts), federal learning, and cutting -edge research on ZKML, Allora has unlocked huge new application design space in the cross -area of cryptocurrencies and AI.
Follow the reason: Allora was developed by UPSHOT. UPSHOT has been a market leader who has developed AI X encrypted infrastructure in the past 2.5 years.They focus on more financial use cases: AI -driven price information flow, AI -driven DEFI vault, AI risk modeling, etc., which may mean that they find PMF earlier than most people.
How to get positions: Join Discord and pay attention to how to participate as early community members.
9. BOTTO
Introduction: Botto is a completely autonomous artist with a closed -loop process and output that is not changed by hand.The only input of human beings is to vote on the BOTTO output to guide the artist what to do next.
Fantastic reason: This unique project combines AI, art, NFT, and DEFI, and has generated actual income (since its establishment is $ 4.5 million).Botto’s artwork has been sold in Christie’s Auction Fair.This is the first AI artist who can invest in history.Art sales revenue will be assigned to pledges.
How to get positions: Botto tokens or buy BOTTO NFT on Super Rare.
10. Parallel (Colony)
Introduction: Colony is an endless game, driven by AI, all simulation items are on the chain.You will be paired with a Parallel.You and your incarnation will work together and share resources on the chain to control the expanding Parallel world driven by Prime.
Follow the reason: Prime is one of the only tokens that are really intersecting the game and AI.”Colony” may become a new type of game. If the team is performed, it has real virus propagation potential.The studio produced this game may be the best in the field of web3 games.
How to get positions: Prime tokens and Parallel avatar NFT.Please register to play the game when the game is released.
11. AETHIR
Introduction: AETHIR introduces a new cloud computing infrastructure method, focusing on the ownership, allocation and use of enterprise GPUs.It acts as a market and polymer to promote the connection between users and organizations in intensive industries such as AI, virtualization computing, cloud game, and cryptocurrency mining.
Optimistic reason: AETHIR looks like another powerful Depin competitor in the GPU computing cloud category.The number they claims 20 times more GPU than Render.They will launch in a very favorable environment in the popular industry.
How to get positions: The upcoming node is sold and joined their Discord server.
12. Morpheus
Introduction: Morpheus is building the first truly decentralized point -to -point personal smart network to achieve the popularization of AI.
Fighting reason: A cool fact about this project is that one of its contributors is Erik Voorhees, who is the real OG in this field.This project gives me the feeling of Bittersor.
How to get positions: You can invest STETH during the fair launch to earn the MOR tokens
13. Autonolas
Introduction: Autonolas is an open market for creating and using decentralized AI intelligence.But not only that, it also provides developers with a set of tools to build AI intelligence under the chain, and can insert multiple blockchain, including Polygon, Ethereum, GNOSIS Chain or Solana.
Optimistic reason: Autonolas is one of the few evidence that has been used one of the AI projects that have been adopted.OLAS is one of the minority tokens that are currently bidded in artificial intelligence encryption projects.
How to get positions: Olas tokens
14. MyShell
Introduction: MyShell is a decentralized comprehensive platform that is used to discover, create and pledge AI native applications.
Fantastic reason: MyShell is an AI application store and a platform that allows you to create AI robots and applications.It allows anyone to become AI entrepreneurs and monetize through their applications.This product is now put into production.
How to get positions: Although they have no tokens, you can register their applications and start interacting with robots to earn points (who knows what it will bring to you).
15. Origintrail
Introduction: Origintrail integrates blockchain and AI, providing decentralized knowledge graphs (DKG) to ensure the integrity and source of data, and enhance AI functions by providing access to verified information networks.The consolidation aims to improve the efficiency and reliability of AI intelligence in various industries through the establishment of a security and trustworthy foundation for data creation, verification and query.
Fantastic reason: the product has been running.Corporate customers.My understanding is that the knowledge map allows AI to explain data and understand it in the context of other things.Trac seems to have a batch of fanatical followers.
How to get positions: TRAC tokens
16. Ritual
Introduction: Ritual is an open, sovereign AI execution layer.Ritual allows developers to seamlessly integrate AI into applications or protocols on any chain, so that they can use the encryption solution to fine -tuned, monetize, and inference the model.
Ritual’s vision is to enable developers to build completely transparent DEFI, self -improved blockchain, independent intelligence, generating content, etc.
Optimistic reason: Ritual does have top supporters.Developers can try Infernet SDK now.I found that a developer used the SDK to start an experimental NFT project a few days ago.Very cool (I am too late, I have not had time to cast).
How to get positions: Add their Discord and continue to pay attention.
17. Nillion
Introduction: Nillion can train and reasonable AI models in a safe and confidential way to create a pillar of safe and personalized AI.
Following the reason: Nillion’s blind calculation network has unlocked many new use cases, of which personalized AI is a huge unlocking field.Unless there is a private data processing, personalized AI will not be widely used.Nillion’s solution sounds changing the rules of the game.
How to get positions: join their discord and keep tracking.If you are a developer, I believe they will host some hackers soon.
