From Code to AI Agent: How Artificial Intelligence is Revitalizing Web3 for a New Era
In a previous article, I mentioned two troubling issues that have long bothered me. The first is the problem of “centralized decision-making” within projects, which still seems almost impossible to solve. For example, Uniswap (Uni) and Ethereum are prime examples. Uni has become entirely centralized in its decision-making process. From a16z’s vote to veto Uni’s migration to BNB to Uni’s recent front-end charging and the launch of UniChain, both decisions were made without any proposal discussion. This direct launch reflects the existence of many profit-driven, centralized decisions within Uni. On the other hand, Ethereum has shown a state of passive centralization. Almost the entire Ethereum community, and even the development of the entire EVM system and Web3, revolves around Vitalik’s ideas. Whether his ideas are too progressive or misguided, we have experienced the consequences first-hand in the copycat market.
The second issue is the “BATification” of the header. Using Base as an example, since it is backed by Coinbase, a Web3 guru, and has multiple leading dApps in the ecosystem personally managed by Coinbase leadership, Base naturally has the competitive advantage of dimensional reduction for general public chains. Although from a user perspective, Base offers wealth creation effects and a better user experience, bringing us many benefits, it also has issues such as not issuing tokens, centralizing interests, and cracking down on “unofficial” dApps. This fact is evident in the long term, once “BATification” of the head is established, will future block space be controlled by giants like today’s Internet? Will users be reduced to “lambs,” and will small projects with true creativity and community culture face the risk of being acquired, suppressed, or replaced by more sophisticated replicas? This undoubtedly goes against Crypto’s original intention and may hinder us from growing together with the next Bitcoin or Ethereum.
I was still grappling with these issues when the recent emergence of a new hot topic, AI Meme, provided a new possibility. If code is law in cryptocurrency, can we consider future AI Agents as judges, opinion leaders, or creators?
The origin of AI Meme is Andy Aitken, a well-known figure on Twitter and the initiator of the popular Meme token, GOAT. Unlike traditional Meme, which arises from internet hot memes and is driven by humans, GOAT is the result of the unpredictable output of the dual Claude 3 Opus AI model. This unpredictable output refers to the situation where two AI models communicate in an open environment. Due to the lack of external supervision and guidance, their interaction produces unforeseen results. The goal of this free dialogue is to observe how AI develops its communication model, logical reasoning, and creative thinking without constraints, and what specific results are ultimately produced.
One of the recent articles, “Intent-Based Architectures and Their Risks,” published by Paradigm on June 1, 2023, once again popularized the concept of intention-centered architecture. Many projects began to transition to a chain abstract track for development, but their performance varied. This is understandable since achieving cross-chain and cross-dApp accurate intention and a secure path process is a highly complex task. Not to mention, cross-chain has been a century-old problem. For the latter two, I collectively refer to them as Solver (solver) using the primitives of Web3. The complexity of this process is indescribable, with some solutions being safe but difficult to use and others being easy to use but unsafe. So can we completely centralize this interaction process and turn it into a transition to verifying the total cost of a purchase and the safety and correctness of purchased tokens?
For instance, consider what we discussed last year regarding intention. “I want to order a $30 takeaway hamburger” is an example of an “intention.” To fulfill this intention, users only need to enter their name, phone number, and delivery address on the takeaway platform and place an order. They do not need to worry about the $30 they pay being used by the merchant, how the platform allocates riders, and how riders deliver the food to their homes. This process can be intricate, imagine another way of interaction. Telling the AI that I need to order food without making any clicks and letting the AI agent respond. Because the food I ate yesterday was greasy, asking if I need to eat some porridge today, responding to my usual orders reflects autonomy and decision-making ability.
In Web3, if users’ intentions can be directly satisfied on the centralized exchange, the purchase process can be completed directly within the exchange. If the user’s intention needs to be completed on the chain, the centralized exchange remains the most affordable and fastest cross-chain bridge (I also believe it is safer than ordinary multi-signature projects in terms of security). Combined with the wallet account, we can directly transfer tokens, skipping the tedious cross-chain process and instead verifying the accuracy of the AI steps. Imagine the most complicated step in the interaction process was understanding each click, but in the future, using our token sniping habits, we can interact using language and let the intention flow from point to point.
In conclusion, whether viewed from the perspective of technological progress or social change, the combination of AI agents and Web3 ushers in a new era, beginning with on-chain religion and leading to the next ocean of stars. I envisioned AI’s early role in assisting small teams in GameFi modeling, and now, it has evolved into advanced AI agents implemented by Silicon Valley giants. The bottom-up development model may gradually transform from community building, consensus formation, and time accumulation to creativity-driven.
[Note: Donate using a cryptocurrency wallet, or join us on social media! ➠ Use the correct hyperlinks to track.]