Translation: Simple Blockchain

For a long time, cryptocurrencies have been criticized for their poor user experience (UX) and high operational risks. But what if this "inhuman" design is not a flaw, but rather a form of advanced evolution? This dialogue explores a forward-thinking viewpoint: that blockchain might not have been designed for humans from the very beginning, but rather for artificial intelligence agents.

While humans are still shocked by issues like malicious transactions, private key storage, and blind signing of contracts, AI agents navigate the world of code effortlessly. They are not fatigued, not fearful, and are naturally proficient in machine language. With cutting-edge experiments like OpenClaw, we are entering a new era of dual-track pillows—where humans step back from decision-making, and AI moves rapidly across the barren landscape of the chain. This is not just a merger of technology; it represents a transfer of financial sovereignty from "primate encyclopedias" to "digital brains".

Choosing the Wrong Users: Why is Cryptocurrency Inherently "Inhuman"?

Host: In what ways do AI agents have a comparative advantage over humans?

Hib:The most obvious answer is that it is impossible to execute laws against an artificial intelligence agent. If you are a fully autonomous agent, there is no possibility of violent monopoly. You cannot imprison an AI agent.

Host: Hib, I would like to ask a question: Why does cryptocurrency seem not to be designed for humans? Even as a user of cryptocurrency for 10 years, I still feel fear each time I sign off on a large transaction. I am reflecting on the fact that I have never been afraid of transferring a wire.

Hib:I never worry that if I don't repeatedly check the wire transfer, I might accidentally send money to North Korea.

Host:Right. But I think like this every time I sign a large crypto transaction. The reality is that the cryptocurrency world is filled with "foot guns": when reading addresses, I have to consider whether it’s a poisoned address attack; I need to check the middle digits only looking at the ends; whether any transitional approvals are outdated; and I have to check the URL to confirm it's not a slightly modified phishing site. There aren't that many traps in traditional financial systems.

Currently, the narrative in the crypto circle is that all of this is due to humans being too lazy. Humans are supposed to focus more on security and develop better operating habits. This is the user's problem, not the technology's fault. But the more I think about it, the more I feel that if we are still deceiving ourselves like this in 10 years, the problem might not be with the users but with choosing the wrong users.

Smart Contracts and AI: The Perfect Habitat for Textual Beings

Hib:What truly began to enlighten me was how powerful my AI agents are at handling code, while humans struggle with issues that form the foundation of its deficiencies. I remember that in my first blog when I started out, I mentioned: smart contracts are supposed to replace laws and traditional contracts, hence the term "smart contracts". In the future, one would not need to find a lawyer to sign an agreement, as it would be signed directly by code.

But the reality is that this story has not materialized. We have not replaced legal agreements with smart contracts. In fact, as a crypto VC, when we need to purchase Tokens from foundations or projects, we still sign legal contracts. Even with smart contracts in place, we additionally need to sign a legal contract just in case.

Host:So this implies that these things are not designed for humans, but they are very suitable for non-human participants. You mentioned a metaphor at ETH Denver: most of the people who speak about "smart contracts perfectly replacing traditional law and property rights" are autistic software engineers—the very ones who built Ethereum. However, the majority of Ethereum users are not autistic software engineers. Yet AI agents are still those engineers out there.

Hib:Therefore, it becomes apparent that negotiating a smart contract, analyzing it line by line, finding all possible errors, and even formally verifying it before deciding whether to agree—these are tasks that a model like Claude can accomplish. Whereas humans need to hire software engineers, spend time reviewing code boundaries, think through scenarios, and conduct risk analyses with lawyers. My threshold for smart contracts is far lower than that for legal contracts. But AI agents are exactly the opposite: they find smart contracts much less comfortable than legal contracts.

Host:You mentioned in your blog that legal contracts are actually full of randomness. For instance, when signing a legal contract, you do not know in which jurisdiction it will ultimately be enforced. It might be California, it might be New York, where jurisdictional issues may arise. The terms agreed upon in New York may also be declared invalid. Who are the lawyers? Who is the judge? The judge and jury are randomly selected. These things are designed to be random and uncertain. AI agents view legal contracts as: this is unexplainable, and uncertain.

Hib:Smart contracts are merely machine code, compiled into EVM bytecode, which can be analyzed step by step, and in 100% of scenarios, will yield the same results. While humans may rationally understand this, intuitively they do not perceive it this way. We rather think of legal contracts as more predictable, despite their randomness. This is due to our bounded rationality, as our ability to handle code is not as strong as that of AI agents. But for AI agents, the promises that cryptocurrency made originally—better enforcement and better property rights—are genuinely established.

Host:So your point is that the original promise of cryptocurrency is not fulfilled by humans but rather by artificial intelligence agents on behalf of humans.

Host:I recently downloaded MetaMask at ETH Denver to sign in. Are you still downloading MetaMask? However, I was pleasantly surprised by the UX improvements in MetaMask, which represent progress in the industry. We have indeed been working on improving the experience for human users over the years.

Hib:What you are saying goes deeper than simple user experience improvements. Artificial intelligence is not just helping human users fix the flaws in the cryptocurrency experience. For example, in opening ledgers with blind signatures, AI can parse code and know how to support or oppose. This can improve the user experience for cryptocurrency, but the more profound issue is that blockchain is fundamentally not a technology optimized for humans.

Host:Right, ultimately it serves humans, as value ultimately flows to humans. But is the correct use of humans really to move the mouse, click plugins, input passwords, manually click buttons, and approve gas? This is too counterintuitive for humans and completely contradicts our understanding of money and finance. It’s like if the banking system required humans to write SWIFT codes themselves. SWIFT is a protocol for interbank communication, not designed for humans. To use it oneself, while possible, clearly does not align with human instincts regarding the use of money.

Hib:So my point is that currently, humans are directly interacting with machines, which is leading to a lack of efficiency. This is quite terrible. Just like with cars: in 10 years we will look back in horror and realize that we thought it was a good idea to let primates operate two-ton machines with manual commands on highways while potentially being drunk or fatigued. This would lead to a significant restriction on human driving, or allowed only in specific areas.

Cryptocurrency has also reached this point. We will look back and think: humans actually manually blind sign transactions, and check addresses with their own eyes. Manually looking at URLs to judge whether they are phishing defenses. Humans can make mistakes, get fatigued, and lack the energy to check three times, check DNS, or check Twitter to see if the protocol has been hacked. We don’t have mechanisms in place to alert us automatically when protocols are hacked; we have to browse Twitter and hope to see it. Mistakes will happen. But AI agents never get tired, never slack off, never skip steps, and strictly execute instructions.

Dual-Track Tool: From Manual Interaction to Automated Future of AI Agents

Host:You can imagine a world fully participated in by AI. You tell the AI: "I feel interest rates are going to rise, so I should shift to safer DeFi." The AI automatically executes: moves you from a high position to a low-risk strategy. If you want confirmation, it provides the plan: "This is my plan, please approve." In the near future, it might be approving the plan, and in the distant future, it might be executing directly because humans cannot add any value.

Hib:In this world, you no longer click protocol logos, you no longer look at marketing, and you don’t even specify which protocol to enter. You simply say "lower risk reconstruction path," and the AI filters protocols, checks TVL, singularity, and selects the best one to execute. What about marketing and network effects? Many protocols' business models are built on the surface behaviors of humans: humans will look at the first few, and inevitably select the largest. But AI agents will not think this way.

If this story holds, the way protocols work and compete will change. Ultimately, consumers will benefit the most. Efficiency will be captured by users, beneficial for good users and for cryptocurrency. But this will not happen next; it will come gradually with model improvements.

Host:If cryptocurrency is not designed for humans but for AI agents, then it is crucial to learn to view the world from the perspective of AI agents. There is a book called "Seeing Like a State," which discusses how states view the world. It is difficult to step out of the human perspective. We view UIs, and cryptocurrencies with human eyes. But if we start seeing through the lens of AI agents, we would be better positioned to predict the future. This is a key skill for builders, VCs, and investors.

The OpenClaw project has first shown me how an unfettered AI agent perceives the world. It prefers command lines. Providing it raw data and root access, rather than through an API or wrapped UI, will be much faster. OpenClaw has always wanted to bypass the MetaMask UI, directly take seeds, extract private keys, write transactions with code, skip those flashy UIs designed for humans.

Hib:What you're saying is profound. AI's innovation comes from large language models (LLM), trained on vast texts. Text is central. It is now moving into images and videos, but text remains the strongest. When AI operates a computer, if you give it a screenshot, it will tokenize, but it is fundamentally a textual being. Text contains the linguistic data of entire human history, whereas the training data for computer screenshots is minimal. Interfaces are designed for humans, but the models grow within texts. Text is a highly compressed representation, making it easier for them to learn.

Host:Yes, the most severe UX panic in cryptocurrency occurred when it was all in the terminal. The earliest Bitcoin and Ethereum transactions were executed in command lines. From the beginning, cryptocurrency existed in a form highly compatible with AI. Our poor UX is their "good UX." Google OAuth wallets are actually harder for AI to use. You wouldn’t want AI to possess GoogleTokens, as that grants access to Google accounts. You want it to only hold a pen encryption in an isolated wallet, governed by noisy rules. Cryptocurrency has always existed with UX that AI can perfectly parse.

Hib:The current issue is that AI has yet to be trained to use cryptography. Most are trained in coding, mathematics, dialogue, and similar fields. Recently, OpenAI released EVM Bench, and Anthropic has also published papers demonstrating model attacks on EVM to showcase their intelligence. But most of the time, they are testing generalization capabilities, not training on these specifics. Once they believe crypto is the future mainstream payment method, real artificial intelligence will emerge.

Host:At present, cryptocurrency remains a relatively undeveloped area for AI training compared to other fields.

Hib:This is how it goes for everything that hasn't been optimized. For instance: Claude is not great in chess because they haven't trained on it. They haven't encrypted laser array patterns, partly due to controversies surrounding cryptography and partly due to liability concerns. If it were publicly stated that training the model helps chess users with encryption, and someone messed up, it would definitely make headlines. Even signing disclaimers might not help, as a bad experience would propagate. Risk and returns, etc.

Host:So you believe the main reason they haven’t ventured into this is legal liability. If Claude messes up a transaction and incurs losses, they bear substantial responsibility, and hence hesitate to train openly.

Hib:Absolutely. Unlike coding or medical advice, where risk and rewards differ. Cryptocurrency wallets involve financial operations, which carry entirely different risks.

Host:This is also why OpenClaw is so exciting for the cryptocurrency community: it is not a large corporation, which means there isn't the pressure of legal liability. It's an open-source project, and users assume the risks. No one can sue a third party, allowing it to take these risks. What is the timeline for this AI agent economy?

Hib:Globally, only about 12% of people have used AI products, with the majority being zero users. Among those who have used them, only 1% have paid for them. The diffusion of technology is slower than expected.

Host:Among the 1% who have paid, OpenClaw is again leading the way.

Hib:Yes. After OpenAI acquired OpenClaw, Sam Altman stated that this will be central to future products. However, OpenAI's path differs from OpenClaw. OpenClaw is an open-source experiment, like early automobiles without seatbelts. OpenAI prioritizes safety: there are commercial processes, and buying something requires manual approval. OpenAI probably won't act like OpenClaw for at least five years, as the legal liabilities are too heavy. Visa also doesn’t allow that: if AI buys haphazardly, Visa will reject the refund for actions not done by a human. They will require verification that you are human. Visa is designed for person-to-person interactions; in a world of artificial intelligence agents, economic mechanisms need to change.

Host:So we have a quota track: one track is the world recognized by humans, staying here long-term with safety as the priority. The other track is a futuristic world reminiscent of OpenClaw. They will use stablecoin wallets to pay each other without worrying about 3DS or refunds. Errors by AI will simply become a business cost.

Hib:They will be long-term participants in the alternate track world. The forefront will establish fully automated businesses on-chain. Currently, the models are not good enough, but Claude 4.6 can perform 14 hours of human tasks continuously and is growing exponentially. When the capabilities reach an infinite length, all intuitions will collapse.

Host:If the occupying tracks are valid, the rate of AI adoption of cryptocurrency will lead the success of these tracks. The OpenClaw world can be seen as the earlier phase of the internet.

Hib:If you observe cryptocurrency itself, you can see this. In 2017, Coinbase only listed a few coins as a means of protecting users. The real cutting edge is on-chain: the Arctic, hackers, carpet pulls. Only recently has the Coinbase App started directly supporting Uniswap. It took a long time to feel safe enough. AI is in the same situation: the cutting edge is in the OpenClaw world. Agents will make mistakes, will be subject to illusions. But as training progresses, the error rates will rise.

Host:How can we get AI developers to respect the potential of cryptocurrency instead of only seeing speculation?

Hib:Many who believe in AI also believe in cryptocurrency: Elon Musk, Sam Altman, Zuckerberg. Cryptocurrency does indeed have its controversies and hassles, but it will not disappear. It is like spam in email; it’s rampant but Gmail blocks it for you. AI will do the same: it will block the bad and amplify the good. Technology is never a hybrid. Information is digitalized, and money is also being digitalized—it will not reverse. In the long term, controversy will give rise to bald adoption.

Host:Final question: Did Dragonfly's new fund of $650 million and AI impact your strategy?

Hib:We are heavily looking into this space. Although it's still early, the direction of value flow is still being tested. Personally, I am looking at AI, but we are also observing stablecoins, payments, and DeFi. AI agents represent general intelligence, using what we use, or commanding lines. There may not be that many investment projects specifically for AI. If you're trusting the theory of AI agents, what to buy? It's like when China lifted its ban on cryptocurrency—everything rose. Demand increases, raising the floor time. This is beneficial for cryptocurrency as a whole.

Host:Thank you. While there are risks in cryptocurrency, we are moving towards the forefront of artificial intelligence. I'm glad you're on this unbanked journey. Thank you!

This article link: https://www.hellobtc.com/kp/du/03/6242.html

Source: https://www.youtube.com/watch?v=RTJjXZZN1jg