Chainfeeds Summary:
Perhaps the endless ability to tokenize various underlying assets is the charm of Crypto.
Source:
https://mp.weixin.qq.com/s/X26-yJeqsiwTF1QYKm6j9w
Author:
Zuoye
Viewpoint:
Zuoye: In a nutshell, AI Agent is a specialization of LLM. The current LLM is not AGI, that is, it is not the L5 organizer envisioned by OpenAI, and its capabilities are greatly limited, such as easily hallucinating when consuming too much user input information. One important reason is the training mechanism. For example, if you repeatedly tell GPT that 1+1=3, then there is a certain probability that when you ask 1+1+1=? in the subsequent interaction, the answer may be 4. Because at this time, GPT's feedback comes entirely from the user's personal information. If the model is not connected to the network, it is very likely that its operating mechanism will be changed by your information, and it will become a stupid GPT that only knows 1+1=3 in the future. However, if the model is allowed to be connected to the network, then GPT's feedback mechanism will be more diverse, after all, the majority on the network believe that 1+1=2. To continue to increase the difficulty, if we must use LLM locally, how can we avoid such problems? A simple and rough approach is to use two LLMs at the same time, and stipulate that each time a question is answered, the two LLMs must verify each other, in order to reduce the probability of errors. If that's still not enough, there are other methods, such as letting two users process a process each time, one responsible for asking, and the other responsible for micro-adjusting the question, in order to make the language more standardized and rational. Of course, being connected to the network does not completely avoid the problem, such as when the LLM retrieves the answer from the stupid bar, which may be even worse. But avoiding these materials will lead to a decrease in the available data volume, so we can completely disassemble and reorganize the existing data, or even produce some new data based on the old data, in order to make the answer more reliable. This is the natural language understanding of Retrieval-Augmented Generation (RAG). Humans and machines need to understand each other. If we let multiple LLMs understand and collaborate with each other, essentially we are already touching on the operating mode of AI Agent, that is, human agents calling on other resources, including large models and other Agents. In a sense, this is also the consistent style of Web3 in recent times, that is, the token issuance platform is more valuable than the token itself. Pump.Fun/Hyperliquid are like this. Originally, the Agent should be the application and the asset, but the Agent issuance framework has become the hottest product. This is also a kind of value anchoring thinking. Since there is no distinction between various Agents, the Agent framework is more stable, and it can generate the value vortex effect of asset issuance. This is the 1.0 version of the combination of Crypto and AI Agent. And the 2.0 version is emerging, a typical example is the combination of DeFi and AI Agent. The concept of DeFAI is of course a market behavior stimulated by heat, but if we consider the following situations, we will find something different: Morpho is challenging old lending products like Aave; Hyperliquid is replacing the on-chain derivatives of dYdX, and even challenging the CEX listing effect of Binance; Stablecoins are becoming payment tools for off-chain scenarios. It is against the background of the evolution of DeFi that AI is improving the basic logic of DeFi. If the previous DeFi's biggest logic was to verify the feasibility of smart contracts, then AI Agent is making the manufacturing logic of DeFi change. You don't need to understand DeFi to create DeFi products, which is a further abstraction of the underlying empowerment beyond the chain.
Source
