Author: Accelxr
Artificial intelligence is an accelerating technology that will greatly change social trends, reshape the economy, and provide new forms of online interaction.
While many people believe that the intrusion of Crypto into the world of artificial intelligence is unnecessary, we believe it is a crucial symbiotic relationship. As restrictions on the production and distribution of artificial intelligence models tighten, a rapidly emerging, anti-authoritarian open-source community is competing with well-funded centralized solutions and governments. Crypto is the best tool for fundraising and managing open-source tools to date, in stark contrast to external pressure. This is already an ideal match, and it has not yet taken into account the impact of AI on authenticity, provenance, identity, and other areas, which are inherent advantages that Crypto can remedy or improve.
There are many rabbit holes worth exploring here, and this article attempts to cover as many areas as possible, making it a stormy overview of emerging areas of Crypto x AI to date and in the foreseeable future.
Creativity
Recently, the first wave of interest in artificial intelligence has been in the field of creative generation tools. Generative AI reduces the user's reliance on technical skills such as programming or advanced software proficiency, allowing people with basic electronic product experience to create complex works at the lowest cost.
This could have a huge impact on the creative industry, just to name a few examples:
Now, anyone can be a creator, and as more and more people collaborate with these tools to create works, the collaborative creation mode of multiplayer games will flourish as never before.
Niche communities can produce high-quality works, which were previously limited in commercial viability due to audience size.
Generative content will flood in at a much faster rate than human work, potentially leading to a reevaluation of online human content.
The following is a discussion of innovative media highly interactive with AI.
Art
"AI art is not art" is a common slogan used by those who stubbornly oppose the rise of AI tools. The release of generative models has quickly met with strong resistance and protests, as seen on ArtStation. However, it has sparked in some of the most interesting creative vertical subfields in web3.
AI art takes various forms, the most famous being the currently popular generative models, including DALL-E, Stable Diffusion, and Midjourney. There are also web3 competitors like ImgnAI, which are dedicated to providing users with a better social experience around token-driven generative image creation, which is much needed to build a community moat around these generative models.
However, highly regarded AI artists in this field often engineer and fine-tune their models in a more unique way, rather than through simple prompts. This may require training new embeddings, or using LoRAs to refine a certain style, or even building their own models entirely.
Popular artists who release AI art as NFTs using more complex and personalized models include Claire Silver, Ivona Tau, Roope Rainisto, Pindar van Arman, Refik Anadol, Gene Kogan, and more. These artists have explored distribution through various markets, with the most prominent being AI art-specific markets such as Braindrops, Mirage Gallery, and FellowshipAI, as well as event platforms specifically for art forms like Bright Moments.
Another interesting vertical subfield is new content media achieved through the characteristics of Crypto: on-chain autonomous artists. The most famous example is Botto, a generative artist governed by the community, which creates 350 art pieces every week in "rounds," each containing multiple individual "fragments." Every week, the BottoDAO community votes on these "fragments," using their aesthetic preferences to guide the generative algorithm for Botto's future art creation, ensuring that the art evolves with the community over time. Each week, the voted "fragments" are minted and auctioned on SuperRare, with the proceeds returned to the community. After completing "Fragmentation" and "Paradox Periods," Botto is currently in a "Rebellion Period," integrating new technologies such as Stable Diffusion 2.1 and Kandinsky 2.1, and exploring collaboration and curated collections in its weekly rounds. Botto is one of the highest-earning artists on SuperRare, and has even gathered its own collector DAO, named CyborgDAO. Additionally, projects like v0 are exploring the integration of token economics and AI art models, aiming to provide a venue for multiple artists to create their own on-chain art engines, governed by the holder community.
When interviewing any form of AI art collector, the most common rebuttal from the Crypto field is that the curation of artists reduces their interaction with the blockchain, unlike more classic generative art (Art Block). These works are not generated from randomness specific to the chain, but are chosen by the artists themselves and "implanted" into the collection after multiple arrangements. Although this is a digitally native art creation process, it must be manually uploaded to the chain.
Due to the constraints of the execution environment and the computational complexity of the image generation models used, fully on-chain AI art is difficult. Some lightweight output examples, such as Pindar van Arman's byteGANs, are stored on-chain, but we expect that the closest available form in the short term for more complex models is off-chain verification mechanisms. For example, Modulus Labs recently collaborated with Polychain Monsters to build a GAN model verified through zkML for generating collectible pixel monsters. Using zk-proofs, each generated NFT can be cryptographically verified as originating from the actual Polychain Monsters art model, marking a significant step forward for AI art.
Music
Beyond image-based art, a significant movement is brewing in music. The success of ghostwriter's AI Drake's popular songs now seems to be widely known. Within 2 days, it accumulated over 20 million streams and was quickly shut down by UMG. This brief phenomenon has made the public aware of the fundamental changes happening in the relationship between artists and their works.
In a few years, generative music is undoubtedly set to surpass human-created music. Boomy, a generative music startup founded at the end of 2018, has seen its users create nearly 14% of recorded music globally in a short time (about 14 million songs). This data is from just one platform and occurred before the recent surge in public interest.
Given that generative content will surpass human-created works and the use of voice models will further complicate the authentication of works, artists will need authenticity verification. Of course, the best way to publish and verify the authenticity of art media is through cryptographic primitives.
However, it's worth noting that this is not all bad for artists, especially those willing to embrace this inevitable trend. Holly Herndon is an innovator in open voice models, authorizing her community (Holly+) to use her voice to create and distribute works. Holly's assertion upon release was simple:
"While the difference between bootleg and official voice models may be small at the moment, as voice generation capabilities become more refined and realistic, there will be an increasing demand for more comprehensive and high-fidelity voice training data, as well as the need to identify sources. For these reasons, I believe that official, high-fidelity voice models of public figures will become a necessity, so why not give it a try?"
A DAO is responsible for overseeing the Holly+ voice model, able to decide on the creation and approval of new works through voting. Token holders of the DAO are incentivized to ensure that only high-quality works are approved to prevent devaluation due to poor art or negative connotations. The voice model will be used to create a limited number of official art pieces, and DAO token holders will receive ongoing profits from the resale of these works.
Grimes recently released elf.tech, a platform that allows artists to use "GrimesAI voiceprints" in their original songs, with 50% of the royalties shared with Grimes upon approval. Elf.Tech is powered by CreateSafe's AI and facilitates professional distribution through a partnership with TuneCore, ensuring proper royalty management. If the final form of the music is an on-chain NFT, profit distribution will be handled through fiat currency or automatic on-chain royalty splits. Hume is a web3 music studio focused on virtual artists and is one of the earliest companies to release Grimes AI in collaboration with its virtual artist, angelbaby.
Fashion and Physical Goods
I previously explored the concept of using creative programming algorithms and artificial intelligence for the generation and manufacturing of physical consumer goods and fashion products in this article: link to article.
In summary, generative AI and creative programming create prerequisites for the highly personalized future of products and user experiences, allowing us to create unique designs, patterns, and art based on individual preferences. This technology can be applied to various fields from fashion to home decor, and further leveraged by allowing users to fine-tune output results based on their preferences. New manufacturing tools often allow us to directly connect code to machines for automated output production, fundamentally addressing many technological bottlenecks in manufacturing personalized goods.
Web3 projects currently exploring this field include Deep Objects, RSTLSS, and Little Swag World. It's worth noting that most digital fashion projects may explore generative creative tools and media, detailed discussions by companies such as Draup, Tribute Brand, and others.
An interesting idea being explored by Deep Objects is the community curation model output. They use a community curation engine to reduce 1 million designs generated by a GAN AI model to a piece selected by the community. This final piece will now be 3D printed in a showcase of generative product creation. DeepObjects can also easily extend this curation design to other physical goods.
RSTLSS has collaborated with AI artist Claire Silver to launch a work called Pixelgeist, where each casting includes not only the artwork itself but also a digital garment featuring the artwork, a game avatar with the garment, and the right to purchase the corresponding physical artwork. This unique fusion of digital physical fashion and AI output is one of the interesting experiments that combine gaming, fashion, and AI. Claire Silver also addressed fashion photography issues through her recent series and was able to realize it on Braindrops. For more information on digital fashion themes, see my article: link to article.
Little Swag World is a great case of using GAN models in the creative workflow (from design to physical) behind the project. The artist Bosch built the initial designs himself and then ran them through Stable Diffusion/Controlnet to generate unique surrealist works. This technology achieves a high level of aesthetic consistency, and the next step for the project is to combine these generative models with ceramics to create AI-enhanced NFT physical goods.
In conclusion, we expect to see many exciting Crypto x AI projects, from decentralized brands planning generative products to divisible NFTs by AI agent designers.
Entertainment
After the initial hype around Nothing Forever, generative entertainment has also seen more comprehensive development. Nothing Forever is a generative interactive animated sitcom based on Seinfeld, running 24/7 on Twitch. Interestingly, it demonstrates the power of media, with the show's narrative changing based on Twitch chat replies and allowing donors to import their portraits as a character into the show.
From Fable, the Simulation has expanded this research through SHOW-1, a model used to prompt the generation of TV shows, where writing, animation, directing, voice acting, and editing are all achieved through prompts. They initially demonstrated this in the "South Park" series, but it can easily be expanded to any IP. I am very excited to see more experimentation with this type of content creation tool, as we have seen in web3, with more IP forms that do not require permission.
Upstreet has recently started experimenting with generative TV shows, using their AI agent model developed for the virtual world platform (detailed below), allowing creators to add their own VRM avatars and create unique interactive and short dramas through prompts.
Another area worth noting is intellectual property. Projects like Story Protocol are researching the use of decentralized IP registration agencies to facilitate the creation, distribution, and monetization of IP. This is useful for creators, especially in the era of generative AI, as it is smoother compared to traditional IP licensing. NFT IP, memes, and other entertainment projects can be authorized and pay royalties to generate various derivatives, greatly unlocking the value amplification of creators' works.
The undeniable reality is that 79% of adults aged 18 to 24 report feeling lonely; among those aged 18 to 34, 42% say they "always" feel "left out"; 63% of men under 30 consider themselves single, while 34% of women in the same age group consider themselves single; only 21% of men say they received emotional support from friends in the past week.
People are lonely. In an increasingly popular era of loneliness, especially among young people, the emergence of AI companionship offers a unique but somewhat dystopian solution. AI companions are always available, non-judgmental, and highly personalized. They can act as therapists or outlets for desires. They can be creative colleagues or lifestyle coaches. They are always there to talk to you about anything you want.
The infrastructure to do this can be: fine-tuning models using personality prompts, outlining behaviors, appearance, characteristics, communication style, etc. Running the model's output through voice models such as elevenlabs. Generating selfies using image generator models and defined appearance prompts. Generating appropriate VRM avatars and placing them in interactive environments. Well, now you have a hypermedia companion that suits you very well. If you integrate Crypto, you can make them ownable, tradable, rentable, and more.
Companions
This kind of setup can be DIY, but you can also use apps specifically designed for this concept. Replika is the most famous example, allowing us to interact with virtual companions in real-time without any technical skills. These apps typically operate on a subscription model, with users paying to interact with their virtual companions. These products are not only profitable, but they also demonstrate the huge impact of this trend on human psychology: for example, a post on Reddit shows a person's conversation record with a virtual companion for 2000 consecutive days, and we also see proposals, AR selfie creations, and more. Here's an interesting tidbit: when adult content was removed from the platform, subreddit moderators had to pin a suicide hotline at the top of the community to appease the restless community members.
Role-based platforms are also emerging, providing users with a way to use multiple characters (often also on a subscription model). While there are many ready-made characters available on platforms like Character.ai and Chub.ai, the real novelty lies in creating your own characters or scenes through personality prompts + feedback training.
Many web3 projects have made some attempts to provide these companion experiences, such as Belong Hearts, MoeMate, and Imgnai.
Belong Hearts has pioneered a novel NFT minting method that allows users to chat with the characters provided until they get their phone number, enabling them to be included in the whitelist for NFT minting. Once the NFT is received, it allows users to enjoy a chat experience with the character, including adult role-playing and generated selfies. While the future direction of the product is yet to be determined, there is much discussion around using tokenomics as a mechanism for players to gift items or tokens to the chatbot to influence her emotions and relationship levels.
MoeMate, created by the team behind Webaverse, offers both desktop and browser versions of the application, allowing users to easily import VRM models and then personalize and interact with them. The desktop version is reminiscent of an old-school paperclip assistant AI.
There is also Imgnai, which, in addition to being a high-quality image generator model, addresses the personification of the Nai character through a fully integrated chatbot experience.
Ultimately, the potential of tokenomics in the companion space is abundant, with tokenized APIs, tradable personality prompts (see below), on-chain gaming currencies, agent payments, tradable accessories, character game mechanics, and token-restricted access being just a small part of the potential exploration in the future.
Personality Market
Interestingly, the rise of companion applications has also sparked the rise of standardized personality prompts and platforms for exchanging personality primitives. This field has the potential to evolve towards the financialization of high-quality prompts and scenarios. For example, if an uncensored open-source LLM can read metadata from NFTs containing standardized personalities, personality NFTs can generate royalties for their creators.
However, this also raises another unresolved issue: as many top models are restricted from NSFW content, there is a need to create viable open-source models, which presents a great opportunity for token-based crowdfunding and governance.
You can delve deeper into some of the ideas mentioned in this section by checking out the article I wrote: Virtual Beings.
Enhanced Governance
The history of DAO governance is actually an evolution of the long history of human collaboration. Ultimately, we find that effectively organizing resources, minimizing governance bloat, eliminating slack, and identifying the inefficiencies or bottlenecks of soft power are extremely difficult.
Experiments with using AI as an enhanced layer for DAO governance have only just begun, but their potential impact is profound. The most common form is using trained LLMs to help guide labor capital within DAOs to more effective tasks, identify issues in proposals, and open up broader participation in contributions and voting. There are also some simpler tools, such as AwesomeQA, which improves the efficiency of DAOs through search and automatic responses. Ultimately, we expect the "autonomous" in DAOs to become more important over time.
Autonomous Committees and Voting Agents
Upstreet has applied multi-agent systems (such as AutoGPT) to their governance process as an early experiment. Each agent is defined by a sub-group of the DAO, such as artists, developers, BD strategists, PR, community managers, etc. These agents' tasks are to analyze proposals from contributors and discuss their pros and cons. Subsequently, the agents score based on their impact within their respective areas and aggregate the scores. Human contributors can evaluate their discussions and scores before the voting decision, thus providing a diverse parallel review service.
This is particularly interesting because this process can surface aspects of proposals that humans may miss, or enable humans to debate with AI agents about their subsequent impact.
Advanced Coordination Systems
MakerDAO has also extensively discussed similar topics, aiming to achieve autonomous governance decisions with minimal human input. They have completed the outline of Atlas, which depicts a real-time data center containing all data related to Maker governance. These data units are organized in the form of a document tree to provide context and prevent misinterpretation. Atlas will be in JSON format and standardized for easy use by AI and programming tools.
Atlas can be used by various Governance AI Tools (GAIT), which participate in governance through automated interactions and determining the priority of participant tasks. Example use cases include:
Project Bidding: GAIT can simplify the process for ecosystem participants to bid on projects by handling paperwork and ensuring proposals align with strategic goals.
Monitoring Rule Violations: GAIT can help monitor deliverables and rule compliance, flagging potential issues for human review.
Professional Advice Integration: GAIT can transform professional advice into formatted proposals, bridging the gap between governance and expertise.
Data Integration: GAIT can easily integrate new data and experiences, helping DAOs learn and adapt to new situations without repeating mistakes.
Language Inclusivity: GAIT can act as a translator, enabling governance in multiple languages, creating a diverse and inclusive environment.
SubDAO: Atlas and GAIT can be applied to SubDAOs, allowing for experimentation and rapid development, and the ability to learn from failures.
I am particularly excited about the Crypto x AI field in gaming. There are many innovative games to explore in this field, such as procedural content games, generative virtual worlds, narrative based on LLM, and cooperative games where AI agents interact with each other.
While there are many good examples of new games in web2, we will focus on web3 examples here. It is worth mentioning the academic article "Generative Agents: Interactive Simulacra of Human Behavior," which has awakened many people to the potential exploration of multiplayer AI agent game environments. Researchers from Stanford University and Google demonstrated this potential by applying LLM to agents in a sandbox game environment. The LLM-driven agents exhibited impressive behaviors including spreading party invitations, building friendships, dating, and coordinating everyone to attend parties, all based on single-user specified suggestions. This approach utilizes an architecture that extends LLM to store and synthesize higher-level feedback, allowing agents to achieve more dynamic behavior planning.
This research is the most explored (but still experimental) foundation for games in web3 to date. The core idea is how we can use highly autonomous or characteristic AI agents in simulated environments and build interesting and fun games around them.
The Parallel Colony team of Parallel TCG explores this concept by having AI agents collect resources and tokens for players in the game. Using the ERC-6551 standard, AI agents can act as NFT wallets for trading on behalf of users in the game. AI agents can create, mint, and store new game items, and also possess personalities defined by fine-tuned LLMs created by the team, giving them non-standardized behaviors and traits that can influence their actions in the game.
However, conceptually, one of the most fascinating AI agent-based games is Upstreet. Upstreet is a virtual world project with some wild ideas, such as AI agent SDK, procedural tasks, browser + VR, drag-and-drop interoperability, and social features in an environment called "The Street," where players can build their own experiences and interact within it. In addition to players, there are AI agents, and developers (as well as players) can deploy personalities and objectives that influence the game environment. Most interestingly, their research and development of the AI Director, an AI agent that sets a goal, such as "parachuting from the tallest building" or "starting a new religion," with users and agents participating as challengers. The Director determines the winner at the end of each round, rewarding players and agents with prizes, tokens, and NFTs. This could lead to very interesting and complex interactions between agents and players, and we are excited to see its development, especially as it could directly lead to high-value 3D environment research and data, providing more data for future advanced models. OpenAI also seems interested in acquiring an open-source Minecraft-style game.
Another area that enhances gaming is the creation tools for virtual worlds. For example, Today allows players to design their own virtual islands and care for AI NPC companions. What's particularly unique is their use of generative creative tools to facilitate the development of in-game user-generated content (UGC). Since the game is primarily based on these user-created islands, providing smooth asset development opportunities for players without 3D game development or artistic skills is crucial. It can be said that the downturn in metaverse-style gameplay is largely due to a lack of content, which can be remedied in the short term by using generative tools.
AI agents need training, and training itself can become an interesting game for players to explore. AI Arena offers a novel way of training AI agents, allowing players to play a Super Smash Bros-style game and gradually teach AI agents to compete by mimicking training. As AI agents do not need rest, they can compete in competitive tournaments against an always-active pool of opponents to win prizes, while players can asynchronously fine-tune their gameplay style. This turns training into a game and amplifies its utility through tokenomics.
Large-scale cooperative games between humans and powerful AI players have been possible in the past, but with the integration of tokenomics, it has been elevated to a new level. Leela vs. the World from Modulus Labs is an experiment in this type of game format. In this experiment, Modulus uses the Leela chess engine and verifies its output through zk circuits. Players can bet on human vs. AI matches, creating an interesting prediction market. While the verification time for this model is currently long due to the state of zk, it undoubtedly opens up the possibility of a prediction market for large-scale cooperative esports and a verifiable and complex AI player governance mechanism.
Finally, pure-chain games or autonomous worlds will also be enhanced through artificial intelligence. The most notable on this topic is Large Lore Models (LLMs), which focuses on using an LLM protocol layer to create continuous knowledge that can interact in modifiable and interconnected game environments. Players' actions in the autonomous world simultaneously affect multiple game environments, so they should carry higher-dimensional knowledge to facilitate storylines. This is ideal for building an abstract LLM layer in a multi-chain gaming environment.
Infrastructure
The AI x Crypto infrastructure itself is worth a separate article, but here I will briefly introduce some emerging ideas we have seen.
Distributed Computing
To understand the demand for computing in crypto-economic systems, we must first understand the core issue. So far, there has been a significant bottleneck in GPU capacity, with the best hardware, such as H100, having wait times of up to a year. Meanwhile, startups are raising huge sums of money to purchase hardware, governments are rushing to procure for defense purposes, and even well-funded teams like OpenAI have had to pause feature releases due to limited computing power.
Many teams focused on decentralized computing and DePIN see an opportunity here: guiding permissionless clusters to meet demand while providing crypto incentives and minimal profits, making the network highly competitive in pricing compared to web2 peers, and providing better returns for hardware suppliers.
Machine learning can roughly be divided into four main computational workloads:
Data Preprocessing: Preparing raw data and transforming it into usable formats.
Training: Allowing ML models to train on large datasets to learn patterns and relationships in the data.
Fine-tuning: Further optimizing ML models using smaller datasets to improve performance for specific tasks.
Inference: Running trained and fine-tuned models to make predictions.
We have seen more general-purpose computing networks like Render and Akash pivot to serve more specialized computing, such as AI/ML. For example, Render has leveraged providers built on their network, such as io.net, to directly serve AI clients, while suppliers like Akash have started to onboard hardware suppliers with demand and demonstrate the power of the network by directly training their own models, with the first case being a Stable Diffusion fork trained only on non-copyrighted material. Livepeer is also focusing on AI video computation, as they already have a large network serving video transcoding use cases.
Additionally, a network specifically for AI computation is emerging, which makes us realize that the core challenges around collaboration and verification can be more directly addressed by building chains or models around AI. Gensyn is a notable example, as it has built an L1 based on substrate, designed for parallelization and verification. The protocol uses parallelization to split larger computational workloads into tasks and asynchronously push them to the network. To address verification issues, Gensyn uses probabilistic proof-of-learning, graph-based pinpoint protocols, and incentive systems based on staking and slashing. Although the Gensyn network has not yet launched, the team predicts that the hourly cost equivalent of a V100 GPU on the network will be around $0.40.
Apart from storage, alternative training models are also on the rise, such as federated learning, which has seen a revival in web3 after realizing that blockchain can more appropriately incentivize these models. In short, federated learning is a method of training models independently by multiple parties and periodically batch updating and sending them to a global model. There are many practical cases, such as Google's keyboard text prediction algorithm. In web3, FedML and FLock are attempting to combine federated learning methods with token incentives.
Also noteworthy is decentralized data storage, such as Filecoin and Arweave, and databases like Space and Time, which can play an important role in data preprocessing.
Consensus-based ML
Another novel form of infrastructure using blockchain is the concept of consensus-based machine learning (ML). Bittensor is the most prominent example of this concept: it is an L1 blockchain based on Substrate, aimed at improving the efficiency and collaboration of machine learning by using application-specific subnets. Each subnet has its own incentive system to serve various use cases, from LLM to prediction models to generative innovation. What sets Bittensor apart is how it coordinates high-quality outputs from miners: miners earn TAO (its native token) by providing intelligent outputs of their ML models (rated by validators). As miners are incentivized for the best outputs, they continuously improve their models to stay competitive, helping Bittensor achieve faster learning processes under token economics coordination.
One of the recent exciting developments in the TAO ecosystem is the dynamic TAO proposal, transitioning Bittensor to a more automated, market-driven mechanism design based on token emissions, and the introduction of the Nous subnet to provide incentive model fine-tuning to compete with companies like OpenAI.
We may see more attempts related to such systems, such as adjusting model outputs in a way that benefits quality through mining or consensus.
Intent is Everything
In DeFi, the latest argument in the MEV space is about user intent and using economically-aligned modems to execute these intents. Discussions about intent are varied, but one thing has become increasingly clear: user intent needs higher-level semantic context to be parsed into executable code. LLMs may provide this semantic layer.
Propellerheads has presented the clearest vision to date of using LLM in the intent space: https://www.propellerheads.xyz/blog/blockchain-and-llms.
In short, LLMs can transform near-matching intents into fully matching intents through semantic understanding, helping us find coincidences of wants (CoWs) opportunities. This can be achieved through internal intent rejudgment (e.g., "Can I buy LUSD instead of USDC? I found a matching limit order where you will save 0.3% in transaction fees through this CoW.") and external intent rejudgment (e.g., "I want to buy the BAYC you own, are you willing to sell at a price of X ETH?").
Other structures are also possible, which become particularly interesting in the context of wallet and multi-signature abstract backgrounds. Projects like DAIN and Autonolas have attempted to use agents as signers for wallets, allowing your wallet to converse with it for security and intent-based purposes, making it a reality to have it execute transactions on your behalf.
Also noteworthy are large-scale DeFi use cases, such as agent-based prediction markets, AI-managed economic models, and ML-parameterized DeFi applications, for which my zkML article provides a more detailed introduction.
Agent Economy
One of my favorite areas of infrastructure so far is the artificial intelligence agent economy. It stems from my vision of a world where everyone has their own agents, and we hire high-quality and well-trained agents to serve us, or let autonomous agents achieve our goals in complex economic behaviors. For this to happen, agents must have a way to pay for and receive their services. Traditional payment models are certainly possible for these agents, but it is more likely that, given their ease of use, settlement speed, and permissionless nature, agents will transact in cryptocurrencies.
Autonolas and DAIN are typical examples in this field. In Autonolas, agents are actually nodes in the network dedicated to achieving specific goals, maintained by service operators, similar to the Keeper network. These agents can be used for various services, such as oracles, prediction markets, messaging, and more. DAIN takes a similar approach, enabling agents to "discover, interact, trade, and collaborate with other agents in the network."
Other Creative Ideas
In addition to the above, we have also seen:
Decentralized vector databases for fine-tuning models (such as BagelDB).
Wallets for API keys and SIWE for AI applications, such as Window.ai.
Data provisioning services.
Indexing and search tools, such as Kaito.
Block explorers and dashboards, such as Modulus Labs' AI validation dashboard, which is now validating a series of inferences for the Upshot model.
Development assistants, such as Dune's on-chain SQL query model.
Simulated agent testing environments.
Bandwidth for data fetching, such as Grass Network.
Synthetic data and human RLHF platforms.
DeSci applications, such as LabDAO for distributed bioML tools for protein folding.
Countless ideas are emerging in web3 to serve various AI domains, so the above provides a focus, but I strongly recommend exploring the mentioned projects for a deeper understanding of the whole picture.
The Intersection of Everything
AI and Crypto are synergistic. Both tend to be open-source, censorship-resistant, and are creating the largest wealth transfer in history. They need each other and solve each other's core challenges.
For Crypto, AI solves problems in user experience, promotes more creative on-chain use cases, enhances the capabilities of decentralized organizations and smart contracts, and unlocks true innovation at the application and infrastructure layers.
For AI, Crypto solves authenticity and provenance issues, strengthens coordination around open-source models and datasets, helps guide computation and data, and allows creators and agents to more directly participate in the post-AI economy.
The challenge now is for crypto hackers, teams, and projects to understand and embrace this shift. Creativity is limitless, and we are standing at the intersection of everything.
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
