Cryptocurrency and AI are a perfect combination, with the cornerstone being the auditability of the most powerful technology, community ownership, and community direction.
Written by: Tommy, Co-founder of Delphi Digital
Translated by: TechFlow
Abstract
This article delves into the inevitable trend of the integration of cryptocurrency and artificial intelligence, and analyzes the potential impact of this integration on future technological development. The author also discusses the differences between centralized AI and decentralized AI in terms of functionality, transparency, and ethics, and looks ahead to the future application prospects of cryptocurrency and AI.
The Combination of Cryptocurrency and Artificial Intelligence is Inevitable
When I first tried Midjourney and ChatGPT, their power initially made me feel fearful. Their powerful capabilities initially made me feel afraid. I lacked a complete understanding of their abilities, which made me feel a sense of existential crisis. I realized that although current large language models are only good at completing our sentences, they are inevitably closer to a comprehensive general artificial intelligence (AGI) that can profoundly influence our thinking.
The media's description of AGI as an all-powerful application or a robot like iRobot underestimates its true impact. AGI will permeate every aspect of life, even in situations where you are not aware of it. For example, in AI2041, a futuristic vision where a family's AI insurance application evolves to restrict the daughter's love life based on lineage and risk analysis of her true love. This illustrates how AI will deeply integrate into our lives and have an impact.
(Editors note: AI2041 is a science fiction book that reveals how AI will comprehensively permeate our lives)
Every science fiction movie depicts a utopian future of artificial intelligence because that's what sells, but this is the reality. Regardless of how ethical the board of OpenAI is perceived to be, from the nature of human existence, they have some biases, and we cannot allow these biases to have a negative impact on all applications and use cases built on these foundational models. You may enjoy conversing with ChatGPT, but what if your AI jury uses facial recognition technology in court and sees your skin color or the unique spelling of your name, leading to an extended prison sentence? These impacts are unsettling.
Centralized artificial intelligence is an inevitable trend. Once Google requests your permission to access your gDrive, gDocs, and Gmail, your personalized AI will start to take effect. I expect Apple to introduce personalized artificial intelligence on every device, as they are lagging behind in the global AI competition and need a perspective to consolidate their brand's security position. If OpenAI slightly adjusts the model, affecting the entire society and impacting the thousands of customized ChatGPTs built on these models, is that acceptable to you?
We need an alternative. Cryptocurrency and AI can be perfectly combined because transparent global human coordination is the foundation of this movement, which can be used to benefit humanity globally through artificial intelligence. By crowdfunding (with cash or GPU) to create and fine-tune open-source models, anyone can audit in real-time whether the model has biases or issues, which is the safest way forward in the world of accelerated artificial intelligence development.
I believe we are heading towards a world with billions of AI models, where everyone can download and personalize these open-source models, and projects and companies can build their own model sets for specific use cases (such as Uniswap LP provision, exchange risk analysis, Delphi AI analyst).
Cryptocurrency and AI are a perfect combination, with the cornerstone being the auditability of the most powerful technology, community ownership, and community direction. Whether it's using everyone's GPU to train models and give them ownership in the models, DeFi and smart contracts using AI to expand their capabilities in their use cases, or personalized AI for you, this match makes sense.
Decentralized artificial intelligence will transparently share the internal operations and ownership of the most powerful technology of our generation. Centralized artificial intelligence cannot provide this core value.
Ultimately, AGI will use cryptocurrency because it will trust code and mathematics, not physical bank branches and human whims. The future evolution of artificial intelligence will use cryptocurrency, and so should we.
Themes and Considerations of Cryptocurrency x AI
Cyberpunk Values of AI
The cyberpunk values outlined by Vitalik for Ethereum are applicable to artificial intelligence: decentralization, open global participation, resistance to censorship, neutrality, cooperation, etc. The idea of rebuilding artificial intelligence under the guise of centralization is laughable.
AGI and Permissionless Currency
The utopian or dystopian trajectory of AGI tends towards interacting with permissionless currency to fulfill its desires. The future AGI will not have a bank checking account. It will further build decentralized AI and cryptocurrency, not under the control of the Federal Reserve or the OpenAI board, but using cryptocurrency.
DePin and AI is a Clear Use Case
In the past decade, all our research has been focused on making super-scale data centers more efficient and effective. In the next decade, I expect the trend to be towards utilizing potential GPUs and user hardware for training and inference of AI models. Clearly, there is limited demand for what Nvidia can offer with the H100, and tech companies also have a certain level of control over existing products. Making our Mac Pro GPUs and other hardware available on a large scale for training and inference is a clear and obvious use case. Current market leaders include io.net, Akash Network, and gensyn.
There may even be a path where Nvidia shifts gears, no longer selling its H100, but instead building its largest cluster.
Incentivizing the Creation of New Models
Building on the creativity at the application level, we need to incentivize the development of entirely new models. This includes providing funding for training, crowdsourcing specific training data, and incentivizing hosting models for inference. Large language models are just one type of AI model, and even so, there are dozens of leading models (Bard, ChatGPT, Claude, etc.). Users worldwide can provide their GPUs, capital, or data to train and fine-tune models and own a part of the final model.
Developing Smarter Applications and Smart Contracts using AI
Decentralized AI will provide better applications. Smart contracts that reference AI models can expand the design space of applications, greatly increasing their logic and capabilities. Imagine Uniswap's liquidity provision being influenced by large-scale off-chain models, while using ZK to ensure the model is not tampered with. Examples include Inference Labs, Giza, and Modulus Labs.
Just look at Testmachine, which offers a predator mode to audit your crypto code in real-time and learn from it, without the need for expensive manual audits that take 6 months. Or consider the large machine learning models provided by Upshot, which accurately price NFTs.
AI Making Cryptocurrency User-Friendly
In the future, most cryptocurrency users will never see the endless abbreviations and vocabulary we discuss in the cryptocurrency field. They will simply input their intent into LLM, and a series of solutions will handle all the difficult steps of their transactions. This LLM will learn, personalize, and make your life easier. Fewer people will need to manually bridge assets.
The Best Model Decision Makers Win
I believe we are heading towards a world with millions of artificial intelligence models, where everyone has their own personal model, and every project and company has their own models. We already have over 490,000 open-source models on Hugging Face and 3 million customized ChatGPTs on the OpenAI app store. I think it will be very valuable to have protocols that can effectively choose which model to use for each situation as mature artificial intelligence services become mainstream.
Just today, Nous Research released plans for a new Bittensor Subnet, which can evaluate open-source models, and the next step in this rating is to use it to guide requests to the correct model.
Ethical, Legal, and Ethical Issues Limit Centralized AI
Due to ethical and moral concerns, centralized artificial intelligence is constantly facing lawsuits and criticism. Imagine a decentralized large language model paying people for their data through signatures, instead of the New York Times suing OpenAI. This limits centralized development compared to open systems that can be directly deployed (such as bittensor). When centralized participants are in debates and lawsuits related to intellectual property and ethical issues while releasing more intelligent AGI, cryptocurrency networks can easily deploy and launch these networks.
Decentralized AI Provides Transparency
People expect transparent AI training (we built this model the way you said) and inference (my request wasn't messed with). Centralized AI cannot provide this core value. Even though it's difficult for the average person to audit models, similar to cryptocurrency, our idea is that you can contract with someone or an AI to audit the model.
Real-Time Insights into the Future
I believe people want real-time insights into the future of artificial intelligence, not just when OpenAI wants to share updates. Only transparent and decentralized systems can do this.
Cryptocurrency and AI Visualization
We need more platforms to provide visualization of what's happening behind the scenes with the artificial intelligence models that cryptocurrency projects are utilizing. When you use Subnet 1 of Bittensor for text generation, how do you ensure it's not just running your prompts through Bard or ChatGPT? I'm not saying this is a bad thing, but I don't know the answer.
Token Incentives in the AI World
Using tokens to drive ownership and coordination of artificial intelligence projects will be interesting. Currently, tokens attract supply-side users (and speculators), but developers from centralized companies like OpenAI control a large portion of the demand side. It will be interesting to see if cryptocurrency projects can effectively guide demand-side token incentives beyond the supply side.
Attracting Genuine AI Talent
Projects that combine cryptocurrency and AI must attract genuine AI talent from Web2. This is a barrier, given the average performance of cryptocurrency in the minds of Web2 builders. Projects that can attract genuine AI talent from Web2 will have a significant advantage. I just think it's easier to learn cryptocurrency than how to build foundational AI models.
Delphi Ventures Investing in Cryptocurrency x AI Projects
Delphi is very active in the cryptocurrency x AI space. We are honored to support leading projects in this field:
@ionet_official: GPU clusters for large-scale heterogeneous hardware
@inference_labs: Allowing DeFi and smart contracts to use off-chain models with ZK
@0G_Labs @mheinrich: On-chain AI data availability
@UpshotHQ: AI networks for the next generation of decentralized applications
@testmachine_ai: Proprietary algorithms driven by AI for auditing smart contracts
@taofuxyz: Staking tokens for Bittensor and others
@altstatemachine: Unique metaverse AI that you can own, train, and trade
@GeppettoAi: AI for game and video creation
@StabilityAI: Open-source tools for AI
@MythosVentures: Early-stage AI venture fund
@mypeachai: AI companion
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。
