本集简介
双语字幕
仅展示文本字幕，不包含中文音频；想边听边看，请使用 Bayt 播客 App。
Banklessation，今天我们邀请到了Alpin Yukseloglu。
Banklessation, we are here with Alpin Yukseloglu.
他是Paradigm的投资与研究合伙人，也是论文《EVM Bench：面向智能合约安全代理的开放基准》的合著者，该论文与OpenAI合作完成，旨在衡量AI代理检测、修补或利用智能合约漏洞的能力。
He is an investment and research partner at Paradigm, also the co author of a paper titled EVM Bench, an Open Benchmark for smart contract security agents written in collaboration with OpenAI, to measure the ability of AI agents to just detect or patch or exploit smart contract vulnerabilities.
我们将讨论人工智能及其能力将如何影响我们的加密生态系统和智能合约。
We're gonna talk about the way that AI and AI capabilities are going to impact our crypto ecosystem, our smart contracts.
Alpin，欢迎来到Bankless。
Alpin, welcome to Bankless.
嗨。
Hi.
谢谢邀请我。
Thanks for having
我。
me.
我想先提出一个非常重大的问题。
I wanna start off the question with a very big this podcast with a very big question.
我们有多大的风险呢？
How at risk are we Okay.
来自AI的风险？
From AI?
AI在智能合约方面的能力对我们的行业构成了多大的威胁？
How large of a threat does AI smart contract capabilities pose to our industry?
是的。
Yeah.
我的意思是，从长远来看，AI对加密货币将变得极其、极其有利，尤其是在安全方面，因为我们将进入一个更加安全的世界，行业的上限也会大大提高。
I mean, in the long term, it's it's now increasingly clear that AI is gonna be extremely, extremely good for crypto because especially on the security front because we're gonna get to a world where where because everything is much more secure, the ceiling on the industry is much higher.
所以我们的合伙人Matt提到，如果你有一个由夫妻经营的杂货店，因为他们无法看到店内的所有情况，他们的规模就有限制。
So our partner, Matt, talks about how, if you have a grocery store that's run by mom and pop, because they can't see everything in the store, there's a limit to how big they can get.
但一旦你安装了监控摄像头，安全就会像这样提升行业的容量和承载能力。
But the moment you add security cameras in, security has this effect of increasing the capacity, the carrying capacity of of an industry.
我认为在短期内，这取决于我们自己，因为模型正变得极其强大，甚至令人震惊地强大。
I think in the short term, it's up to us because the models are getting extremely good, like strikingly good.
当我们大约六个月前开始开发EVM Bench时，这个基准测试完全由涉及资金被盗的关键漏洞组成，当时的模型只能发现不到20%的漏洞，大约只有12到13个。
When we started working on EVM Bench, which is a benchmark that consists entirely of fund draining critical bugs around six months ago, the models were able to find less than 20% of the bugs, like around 12 to 13.
而在我们开发这个基准测试的过程中，这个数字上升到了超过50%。
And just over the course of while we were working on the benchmark, this number went up to over 50%.
在我起草推文和实际发送Jeep 5.3 Codex发布消息之间，这个数字又跃升至超过70%。
And in between when I drafted the launch tweet and when I had to actually hit send with the release of Jeep 5.3 Codex, it jumped up to over 70%.
因此，这些技术正以惊人的速度发展，我们必须以一种能够防御攻击的方式为行业做好定位。
So these things are just growing at a blistering pace, and it's very important that we position the industry in a way that we can defensively protect against attacks.
但从长远来看，我认为这极大地提升了加密行业的承载能力。
But in the long term, I think it massively increases the carrying capacity of crypto.
是的。
Yeah.
我认为你所说的是，从长远来看，我们将接近实现完美的安全性。
I think what you're saying is in the long term, we get something approaching perfect security.
是的。
Yeah.
目前，我们还没有完美的安全性。
Right now, we do not have perfect security.
让我用另一种方式问你同样的问题。
Let me ask you the same question, but a little bit differently.
假设只有恶意行为者、只有黑帽黑客能使用人工智能能力。
Say only bad actors, only blackhat hacked actors have access to AI capabilities.
在这种情况下，我们的行业有多危险？
In that context, how at risk is our industry?
考虑到人工智能能力的提升，我们的智能合约有多容易被利用？
Like, how exploitable are our smart contracts given the increase in AI capabilities?
是的。
Yeah.
我的意思是，当我们接近超级智能水平时，很难说清楚。
I mean, I think it's really hard to say when we approach superintelligence levels.
我认为在目前阶段，模型已经相当不错了，但还没有超过最优秀的人类审计员。
I do think until we hit the like, right now, the models are quite good, but they're not better than the best human auditors.
事实上，我们早已在加密货币领域面临这种威胁模型——存在极其智能的对抗性行为者，他们不断试图破解所有承载巨额资金的软件。
So we already have existed in crypto under this threat model of extremely intelligent adversarial actors that are constantly trying to break all of our software that with all the money in it.
从这个角度看，加密货币已经相当坚固，但当我们谈论技术跃迁至超级智能时，真的很难判断。
So in that sense, like, crypto is already quite hardened, but it's it's just really hard to know when we talk about sort of a technology inflecting into superintelligence.
这类似于编码能力在过去几年里基本呈线性增长，但在去年12月，它们突破了一个阈值，变得优于普通工程师。
This is very similar to how in coding capabilities were increasing mostly linearly over the last several years, And in December, they crossed some threshold where they were better than sort of the median engineer.
很多事情突然变得清晰起来，大家开始意识到这是一个转折点，一个‘天啊’的时刻。
And a lot of stuff clicked for everyone, and it started becoming this moment and this sort of oh, crap moment.
我认为安全领域也会发生非常类似的情况：目前，安全能力正在快速提升，虽然仍呈线性增长，但尚未超越顶尖人类审计员的水平，所以我们还没感受到威胁。
And I think something very similar will probably happen with security where right now it's, like, increasing pretty rapidly at a at still at a linear clip, but it's not as good as the best human auditors yet, so we don't feel it yet.
它还没有真正打破我们的任何前提假设。
It hasn't actually broken any of our assumptions.
但我非常有信心，再过六到八个月，到今年年底，当出现超人类的AI审计员时，这一切将彻底颠覆我们的所有假设，我们必须重新审视并加固所有承载着近一千亿美元资产的智能合约。
But once we hit in maybe six to eight months, I'm pretty confident at this point by the end of the year, a superhuman AI auditor, this will just completely break all of our assumptions, and we'll need to go back and and make sure that we're hardening all of the contracts that are housing the, what, nearly $100,000,000,000 of assets in crypto.
GALAXY致力于连接数字资产与下一代基础设施，为机构提供端到端的服务。
GALAXY operates where digital assets and next generation infrastructure come together, serving institutions end to end.
在市场方面，GALAXY 是领先的机构平台，提供现货、衍生品、结构性产品、DeFi 借贷、投资银行和融资服务。
On the market side, GALAXY is a leading institutional platform, providing access to spot, derivatives, structured products, DeFi lending, investment banking, and financing.
GALAXY 拥有超过 1600 个交易对手，帮助机构应对市场周期的每个阶段。
With more than 1,600 trading counterparties, GALAXY helps institutions navigate every phase of the market cycle.
该平台还通过主动管理策略和机构级质押及区块链基础设施，支持长期资金配置者。
The platform also supports long term allocators through actively managed strategies and institutional grade staking and blockchain infrastructure.
这种规模是实实在在的。
That scale is real.
GALAXY 平台上的资产超过 120 亿美元，2025 年底平均贷款规模达 180 亿美元，反映出生态系统中深厚的信任。
Galaxy has over $12,000,000,000 in assets on the platform and averaged a $1,800,000,000 loan book in late twenty twenty five, reflecting deep trust across the ecosystem.
除了数字资产，GALAXY 还在为人工智能驱动的未来构建基础设施。
Beyond digital assets, Galaxy is also building infrastructure for an AI powered future.
其 Helios 数据中心园区专为人工智能和高性能计算而建，拥有超过 1.6 吉瓦的获批电力容量，是同类中规模最大的站点之一。
Its Helios Data Center campus is purpose built for AI and high performance computing, with more than 1.6 gigawatts of approved power capacity, making it one of the largest sites of its kind.
从全球市场到为人工智能准备的数据中心，GALAXY 正全方位服务数字资产生态系统。
From global markets to AI ready data centers, Galaxy is serving the digital asset ecosystem end to end.
访问 galaxy.com/banklist 或点击节目说明中的链接，了解更多关于 Galaxy 的信息。
Explore Galaxy at galaxy.com/banklist or click the link in the show notes.
Euphoria 让你只需轻点一下，即可在掌中交易。
Euphoria brings one tap trading to the palm of your hand.
基于 Mega ETH，Euphoria 将实时价格图表投影到一个由方格组成的网格上。
Built on Mega ETH, Euphoria takes real time price charts and projects it over a grid of squares.
你只需轻点那些你认为价格将在未来五到三十秒内进入的方格。
You tap the squares that you think the price will enter in just five to thirty seconds in the future.
如果价格进入该区域，你的交易收益可高达 2 到 100 倍。
If the price goes into that quadrant, you can pocket anywhere between two and a 100 x your trade.
没有其他应用能像 Euphoria 一样，让你在 FOMC 会议、总统演讲或全球宏观事件等市场驱动时刻，以更快的速度和更高的杠杆进行交易。
No other application helps you trade faster and with more leverage on market driving events like FOMC meetings, presidential speeches, or global macro events.
得益于 MegaEth 的实时区块链，Euphoria 是实现与市场实时价格互动的最佳方式。
Thanks to MegaEth's real time blockchain, Euphoria is the way to get real time price interactions with the market.
在 Euphoria 上，你可以通过其实时社交交易体验与朋友竞技，直接与好友一较高下。
On Euphoria, you'll be able to compete with friends using Euphoria's real time social trading experience, allowing you to go head to head with your friends.
如果你把这款应用投射到电视上，这将是一个很棒的派对小把戏。
A great party trick if you project the app on a TV.
这就像衍生品界的马里奥派对。
It'll be like the Mario party of derivatives.
要使用EUPHORIA进行交易，用户可以从任何链存入稳定币，或直接进行法币转账，所有资金都会在后台自动转换为Mega ETH的原生稳定币。
To trade on EUPHORIA, people can deposit stable coins from any chain or do direct fiat transfers, and everything gets converted into Mega ETH's native stable coin in the background.
前往euphoria.finance了解详情，下载应用，或在Telegram中搜索迷你应用。
Check it out at euphoria.finance and download the app or find it in telegram as a mini app.
2024年，新兴市场为投资者创造了超过1150亿美元的年收益，收益率介于10%至40%之间。
In 2024, emerging markets generated over $115,000,000,000 in annual yield for investors, with yields ranging between 10 to 40%.
这些是地球上最高且最持久的收益率之一。
These are some of the highest, most persistent yields on earth.
问题是什么？
The problem?
去中心化金融无法接触到这些机会。
DeFi can't access them.
BRIX 改变了这一切。
BRIX changes this.
基于 Mega ETH，BRIX 将新兴市场的货币市场和套利机会转化为可组合的原语，您可直接从钱包访问。
Built on Mega ETH, BRIX takes emerging market money markets and solvent carry and turns them into composable primitives you can access straight from your wallet.
当 DeFi 投资者在稳定币和国债上获得 3% 到 6% 的收益时，机构却在主权货币政策支持下获取 10% 到 50% 的收益率。
While DeFi investors earned three to 6% on stablecoins and T bills, institutions have been harvesting 10 to 50% yields backed by sovereign monetary policy.
BRIX 通过机构级代币化、本地银行通道、跨司法管辖区的合规性以及实时稳定币结算，连接了这两个世界。
BRIX connects these worlds with institutional grade tokenization, local banking rails, compliance across jurisdictions, and real time stablecoin settlement.
BRIX 承担了繁重的工作，让 DeFi 终于能够接触真实抵押品和基于真实世界收益的结构化产品。
BRIX does the heavy lifting so DeFi can finally access real collateral and structured products on top of real world yield.
即使是最好的套利交易，现在也可能触手可及。
Even the best carry trades can be within reach.
BRIX 将 DeFi 的承诺带入新兴市场，同时将新兴市场的收益带入您的钱包。
BRIX brings DeFi's promise to the emerging world and brings the emerging market yield to your wallet.
让收益通过 BRIX 流动。
Let the yield flow with BRIX.
阿尔平，如果我们把视角拉远，思考一下人工智能的智能及其安全能力和漏洞检测能力正在呈指数级增长，再想象一个超级智能AI，我甚至不知道该如何思考安全问题，因为它能构想出超越人类理解的场景。
Alpin, when we if we zoom out here, though, and we think about AI intelligence and its security capabilities and its bug detection capabilities kind of going exponential and we think about a super intelligent AI, I don't even know how to think about security in general because it can envision scenarios beyond human comprehension.
比如，如果它想出了一种用我们从未知晓的数学方法破解某些加密技术的方式呢？
For instance, what if it thinks up a way to crack some of our cryptography with some math that we didn't even know existed.
对吧？
Right?
我最近其实听贾斯汀·德雷克在一个播客里谈到过这个。
Like, I heard actually Justin Drake on a podcast recently talk about this.
这不仅仅是量子计算机带来的威胁——那是一个真实存在的已知威胁，我们的一些加密算法正因量子计算机而面临风险。
It's not just the threat of quantum computers, which is kind of a real known threat, and some of our, you know, encryption algorithms are under threat due to quantum computers.
但如果我们有了超级智能AI，谁知道它实际上能黑入什么、破解什么。
But if we have a super intelligent AI, I mean, who knows what it could have the ability to actually hack and Mhmm.
解密。
Decrypt.
我的问题是，面对超级智能AI，安全是不是根本就无法提前准备？
I mean, I guess my question is when it comes to super intelligent AI, is security just, like, not even a thing we can prepare for?
我的意思是，我们该怎么去思考这个问题呢？
I mean, how do we even think about it?
我的意思是，我认为安全问题目前的前沿领域非常难以理解。
I I mean, I think security so the way I would think about it is that I think right now, this frontier is very illegible.
如果你试图从极端或极限的角度去思考，最终会陷入一些非常奇怪的境地，甚至可能让人精神崩溃。
And if you try to do this at the limb in the limit thinking, you end up leading to very odd places that that may be, you know, very psychosis inducing.
我认为，具备某种能力是很重要的。
I think one of the skills of Right.
但我觉得，具备某种正确的能力很重要。
But, like, I think I think one of the right.
我认为，面对奇点并保持理智，是一种非常重要的能力。
I think I think the capacity to face the singularity and stay sane is a very important skill to develop.
我认为，我们目前能做的最好的事，就是让自己进入这个前沿领域，进入这种实验性的未来，亲自开展实验，并在关键转折点来临时做好应对准备。
And I think this is you know, the best we can do right now is that we can get ourselves into the frontier, into this sort of experimentally bound future where we're running the experiments ourselves, and be ready when those inflections happen to be able to react.
因为我认为，那种认为世界上只有坏人，他们会获得这项技术并摧毁我们所有系统的观点。
Because I think the the model of, like, there are only bad people in the world, and and they're gonna have access to to this technology, and they're gonna break all of our systems.
我认为这正是导致人们陷入这种疯狂焦虑的原因，比如我们是不是全完了？
I think this is what leads to this sort of psychosis around, like, are we all just completely screwed?
但那并不是会发生的事情。
But that's not what's gonna happen.
对吧？
Right?
我们都会一起置身其中。
We're all gonna be in there together.
因此，当我们身处这个前沿，并能接触到这些前沿模型以及围绕它们构建的、能够运行这些攻击的智能体时——比如发现未被发现的数学定理，或破解现有的图像技术——这些技术将同时具备正反两面。
And so when we're in this frontier and as we have access to these these frontier models, the the sort of agent harnesses around them that are able to run these exploits, that are able to, for example, find undiscovered math or that are able to break existing photography, there will be both sides of it.
而目前还不清楚，这会是一种偏向进攻还是防守的技术。
And right now, it's not clear whether this is gonna be an offense or defense favoring technology.
我想说的是，世界上仍然存在一些根本性的限制。
I will say that there are still fundamental constraints in the world.
比如，你无法违背物理定律。
Like, you can't break laws of physics.
有一些系统是混沌的。
There's there's there are systems that are chaotic.
比如三体问题，即使你拥有超级智能，也无法预测太远的未来，因为这是一个根本性的混沌系统。
Like, for example, the three body problem where, you know, the fact, like, the fact that even if you have superintelligence, you can't predict too many more ticks ahead because it's just a fundamentally chaotic system.
所以我认为，世界是一个复杂的地方，仍然存在物理定律和限制，这些会制约这些事情。
So so I do think that there's this you know, the world is a complicated place, and and there there are still physical laws and constraints that will catch these things.
实际上，我们能做的最好的事情就是共同推动这一前沿，而不是任由它被动地发生在我们身上。
And and practically, the best we can do is we we have to push that frontier together instead of instead of letting it just sort of happen to us.
我觉得这是一个合理的观点。
I think that's a fair point.
这里确实存在物理限制，超级智能AI并不意味着它能像神一样对待人类，更不会赋予它打破宇宙物理定律的神力。
There there are physical constraints here, and super intelligent AI does not just mean, like, it could appear as a god to humans, but it does not give it, you know, godlike capabilities to break physical laws of of the universe, certainly.
是的。
Yeah.
但事实上，我想深入探讨这一点，因为我感觉你可能对此有一些见解，因为我一直在凝视着奇点的深渊，努力保持理智。
But, actually, I wanna dig into this because I I got the sense that maybe you have some insight here because I've been doing a lot of staring into the abyss of the singularity and trying to stay sane.
我对这个并没有什么天赋。
And I I don't quite have a knack for it.
有时候我觉得自己只是盯着看，感觉自己有点要疯了。
Like, sometimes I feel like I'm staring and I am feeling myself go a little insane.
我只是觉得对此无法安心。
Like, I just don't I don't feel settled about it.
你能不能分享一些智慧，比如某种模式？我刚刚从你的话里隐约捕捉到一点东西。
Is there some wisdom you can share just like a pattern or something that like I I was starting to get out of what you just said.
也许关键是一步一步来，嗯。
Well, maybe the key is to kind of take it a step at a time and Mhmm.
不要去想遥远的未来和极限。
Don't think about the far future and the limit.
只需想想明天、下个月、明年。
Just maybe think about the next day, the next month, the next year.
当你凝视奇点时，你是如何保持理智的？
What what's how do you stay sane when you when you stare at the singularity?
我的意思是，核心要点在于能动性。
I I mean, I think the core the core point is agency.
对吧？
Right?
彼得·蒂尔有一个观点，认为接受和否认，大多数人把它们看作对立面。
So Peter Thiel has this framing where acceptance and denial, most people relate to them as opposites.
但在很多方面，它们其实是一回事，因为两者都暗示着你对一切毫无掌控。
But in many ways, they're the same thing because both of them imply that you're sort of everything is out of your control.
如果你完全接受某事会发生，那你也就不会去采取任何行动。
If you're fully accepting that something's gonna happen, then then you're not doing anything about it.
而如果你完全否认某事会发生，你也同样不会去采取任何行动。
And if you're fully denying that something's gonna happen, you're also not doing anything about it.
从这个角度看，我认为末日论者和加速主义者都是错的。
In that sense, I think both the doomers and the the accelerationists are both wrong.
真正的答案是我们对这些结果拥有能动性，你自己就能改变未来的走向。
And the the real answer is is that we have agency over these outcomes and that you yourself can bend the arc of the future.
我认为这种观点让人感到安心且稳定，因为如果你相信自己对结果有掌控力，那么某种程度上，你依然掌握着主动权。
And I think there's there's a lot of comfort in that and a lot of stability in that because if you believe that you have agency over the outcome, then then somehow, like, you're still in control.
我要说的是，当前的前沿领域，或者说，你可以认为前沿始终受到实验的限制，因为我们并不确定。
Now I will say that the current frontier, and, like, I guess you can argue the frontier has always been experimentally bound, which means we don't know.
我们无法坐下来理论化地推测未来会发生什么。
Like, we can't sit down and theorize about what is going to happen.
通常，我们了解事物如何发展的途径是通过实验，然后观察结果。
Generally, the way that we're gonna figure out how things are gonna happen is by is by by experimenting and then by seeing the results.
这些人工智能模型以及当前的技术前沿都是逐渐发展起来的。
And, you know, these AI models and also the current frontier of technology is grown.
它们不是被人为制造出来的。
It's not manufactured.
它们以一种我们今天无法真正预测的方式自然演化而来。
It's much more organically evolved in in a way that, like, we can't today really predict.
因此，仅仅坐在椅子上空想未来会发生什么，就会让人发疯，因为你根本无从知晓。
So even just the the, like, armchair trying to theorize about what's going to happen will drive you insane because you can't know.
它并不受你能否通过算法预测未来的能力所限制。
It's it's sort of a it's not bound by your ability to to algorithmically figure out the future.
所以你必须深入一线，亲自去尝试。
So you have to get in the in the trenches and try things.
是的。
Yeah.
有一种斯多葛式的接受，即你无法知晓一切，因此你只能放下这种执念。
There's some stoic acceptance, the fact that you can't know everything, and so you just have to let that go.
同时，你也相信自己拥有能动性，或者说是一种实践性的信念。
And then there's also a belief in agency and, I guess, an applied belief.
我的意思是，对。
I mean Yeah.
是的。
Yeah.
能动性中有多少是盲目的信仰，又有多少是基于实际行动的务实信仰呢？
How much of how much of agency is just sort of a a blind faith versus maybe, like, a practical faith of doing?
对于那些听着觉得自身正面对奇点、感到冻结、不知所措、缺乏控制感和能动性的人，这种能动性是可以培养的吗？
How do for for someone listening in general that feels themself looking at the singularity and it feels frozen, I guess, by by, like, just like not knowing what to do and doesn't feel like they have a lot of control and agency, is that something that they can develop?
还是你的建议是：你只需要有信念？
Or is your advice, oh, you just have to have faith.
你只需要相信自己拥有更多能动性，这就会自我实现，你最终会拥有更多能动性。
You just have to believe that you have more agency and that will become self fulfilling and you you will have more agency.
我认为信念固然好，但这种心态并不特别有助于激发能动性。
I I I think so faith faith is faith is good, but it's not a particularly agency inducing headspace to be in.
比如，大约八个月前，当我们开始意识到智能合约可能被代理程序极其高效地利用时——而我们显然深受其影响——我们本可以只是坐下来，惊呼：天啊。
I mean, when we, for example, started having the thought that agents could get extremely good at exploiting smart contracts, which we're obviously heavily exposed to, you know, about eight months ago, we could have just sat down and been like, holy crap.
我们完蛋了吗？
Like, are we screwed?
我们没完蛋吗？
Are we not screwed?
是的。
Yes.
是的。
Yes.
对吧？
And like right?
然后结果发现，还有另一条路，那就是去弄清楚这些风险到底有多大。
And and and then there it turned out, like, there's another path, which is, well, you can go figure out to what extent these things are at risk.
同时，也开始着手接触那些真正推动这一前沿的实验室，或许能推动加密技术融入其中，这样我们就能处于一个有利位置——当战争迷雾逐渐散去，当我们开始看到，比如，我们其实有可以采取的防御措施时，我们已经准备好去实施这些方案了。
And and then also start making making headway into the the labs that are actually pushing this frontier and maybe start getting getting crypto integrated into them so that we can get into a position where where, you know, as the fog of war clears and as we start to see, for example, now that that there are defensive measures that we can take, that we're in a position to actually exercise exercise those those paths.
我觉得，那种末日论，以及在极限层面思考奇点和超级智能的思维方式，确实很容易抓住人们的注意力。
And and I think I think there's something about, like, the the doomerism and also the the the general, like, in the limit thinking about about the singularity and about superintelligence that is just, like, captures people.
因为你可以坐下来，花很长时间思考它，它会让你产生强烈的情绪。
Because it's like, you can sit down, you can think about it for a long time, and and it'll make you feel very strong emotions.
但归根结底，那并不会真正让你去动手建设些什么。
But at the end of the day, like, that's not going to be the thing that is actually like, you can go build things.
你可以去和那些推动前沿的人合作。
You can go work with the people who are pushing the frontier.
你可以深入一线，开始贡献力量。
You can get in the trenches, and you can start contributing.
而你从实践中获得的信息，会比任何人在头脑中空想出来的都要扎实得多。
And and the information you gain from that is going to be much more grounded than whatever one can come up with in in their head.
这一点我想很多加密领域的人早就习惯了，因为加密货币的大部分历史都是这样的。
And this is I mean, I think many of us in crypto are used to this because a lot of crypto was like this for most of its history.
它曾经极其难以理解。
It was extremely illegible.
当时很难确切地界定它的应用场景，而我们现在开始看到，比如，我们有了价值存储这个应用场景。
It was very hard to pin down exactly what the use case was going to be, and now we're starting to see, okay, we have this store of value use case.
我们还见证了稳定币以惊人的速度逐年增长。
We had stablecoins sort of compounding at this monstrous rate year over year.
这个行业催生了预测市场这一整个领域，虽然它现在与加密货币关系密切，但其增长速度依然令人难以置信。
The industry kind of gave birth to the whole market of prediction markets, which which is sort of adjacent to crypto now, but it's just compounding at this insane rate.
五年前，很难说清这些事情会具体如何发展，而要真正去构建这些事物，需要一定程度的信念与行动力的结合。
And, you know, five years ago, it would have been very hard to say this is exactly what's gonna happen, and it took some level of some combination of faith and some some combination of agency to to actually go and build those things.
我认为，从文化上讲，这一直是Paradigm的一个重要基石，因为我们的整个公司都是围绕着与投资并行地进行建设和研究而建立的。
And I think so culturally, this has been a very big anchor point for Paradigm because our entire firm is built around around building and researching alongside the investment.
所以，如果你和Paradigm团队或我们圈子中的任何人交谈，你会感受到一种扎根感，这种感觉源于我们始终与前沿保持持续接触，这种接触带来了一种安心、稳定和能动性。
And and so if you talk to anyone on the Paradigm team or in our orbit, like, the the sense of groundedness you get is is anchored in the fact that we're actually we have contact consistent contact with the frontier, and and and there's there's sort of there's comfort in that, there's stability in that, and there's agency that comes from that.
我认为，值得去理解和反思的是，为什么这些直面虚无的‘奇点’令人恐惧，原因在于，所有这些技术实际上都在赋予每个人能动性，让他们能够创造出奇点本身。
I think it is worth understanding and reflecting on, like, the the only reason why these singularities staring into the void is intimidating is because what it's what all of these technologies are doing are providing everyone else with agency to produce the singularity in the first place.
所以，如果这个奇点让你感到害怕，那就拿起拖把，做点什么吧。
And so if that singularity is intimidating to you, you know, grab a mop, you know, get do do something.
有很多工作等着我们去做。
Like, there's work to be done.
我认为，最好的前进道路就是通过这种方式。
And I think, you know, the best path forward is through there.
如果你因为这些工具赋予了他人能动性而感到不安，那你也可以拥有自己的能动性。
Like, if if you are intimidated by everyone else having agency because of these tools, you can have agency yourself.
这是一项具有争议性的技术。
Well, it's a it's a polarizing technology.
对吧？
Right?
你能执行某件事所需的能动性门槛正在降低。
It's it's like the the the threshold of agency you need to be able to execute on something is going down.
所以，如果你犹豫不决，就会被推到零点，或者你可以做很多事情。
So so, like, if you're kind of on the fence, you get snapped to zero or to you can do a lot of things.
或者推到一。
Or or to one.
是的。
Yeah.
我认为，在这种情况下，如果一个人直觉上觉得自己可能处于那条分界线的一侧，被挤压到零，这会让人极度恐惧。
And and I think in that sense, like, if one has the intuitive sense about themselves that they might get they might be on the side of that fence where they get squeezed to zero, that can be extremely fear inducing.
而实际上，解决办法是保持更高的开放性。
And, actually, the solution to that is to is to be much higher openness.
对吧？
Right?
更快地采用这项技术。
Adopt the technology much faster.
在变化和适应环境方面要更加灵活。
Be much more fluid about about about changing and adapting to to the environment.
我认为，马特和我们的团队曾经提到过，有时候速度比凝聚力更重要。
There's, I think, Matt and our team had mentioned at some point about how there are times when speed is more important than cohesion.
我认为，我们当前所处的环境，因为前沿领域如此未知且在某种程度上不可预测，因此快速行动和快速适应比坐下来精确预测未来并精准定位要重要得多。
And I think the current environment we're in, because the frontier is so unknown and there's and so unknowable to some extent, There's there's moving fast and adapting fast has a premium over being able to sit down and figure out exactly what's gonna happen and put the pin in the right place.
这在某种程度上是矛盾的，因为你的自主性越强，就越会感觉选对游戏、选对要玩的项目等事情至关重要。
And this is somewhat paradoxical because the more agency you have, the more it may seem like it matters that you do the game selection right and you sort of pick the right game to play, etcetera.
但实际和经验上发生的情况是，与其花两周时间试图找出构建它的最佳方式再发布，不如在创意诞生后的24小时内快速推进并发布产品。
But practically and empirically, what's happening is that, actually, it's better to move fast and ship the thing within twenty four hours of inception than it is to sit for two weeks and try to figure out the exact right way to construct it and then try to ship it then.
我认为，这种原则可能延伸到个人生活的许多方面，我们现在正处于一个追求速度胜过凝聚力的时代。
And and I think that probably goes all the way down to, like, many parts of of one's life where where we're in an era of speed over cohesion.
是的。
Yeah.
是的。
Yeah.
我们正处于一个只管去做的时代。
We are in the, just do things era.
我原本没料到这个开场会这么富有哲理。
I wasn't Alpin expecting this to be such a philosophical episode to open this episode up.
但现在我觉得我们可以把注意力集中到当前的主题上，那就是是的。
But now I think we can kinda like corral ourselves and point our agency towards the topic at hand which is Yeah.
当人们对AI在智能合约安全性方面拥有高度自主权时，会发生什么？
What happens when people feel high agency with AI towards the security of our smart contracts.
嗯。
Mhmm.
值得讨论一下哪些类型的合约最容易或最不容易受威胁吗？
Is it worth talking about, like, what kind of contracts are, like, most at risk or least at risk?
是否存在某种类别或认知框架，让我们能够理解：当AI对智能合约拥有极高自主权时，我们应该特别关注某些类型的智能合约，而非其他？
Is there some sort of, like, category or knowledge landscape that we can understand that when AIs have very high smart contract agency that we should be paying attention to certain kinds of smart contracts over others?
那里有相关的讨论吗？
Is there a conversation there?
我认为在市场层面很难说清楚，比如去中心化交易所合约和借贷市场之间的区别。
I I think I think not in terms of market, I think that's really hard to say, like, know, a DEX contract versus lending market.
但我认为，那些已经存在很长时间的简单合约，可能比例如在币安智能链上部署的第两百个合约之类的情况更安全，因为过去曾有过一种因市场规模小而形成的保护机制。
But I think, you know, simple contracts that have been around for a long time, I think, are probably better in a better position than, for example, like, the two hundredth contract deployed on Binance Smart Chain or something like that, where, you know, there's in the in the past, there's been a sheltering that's happened from Mhmm.
身处一个小市场中。
Being in a small market.
所以，如果你部署了一个合约，攻击者即使完全利用它，最多也只能获得几千美元的收益，那么你就会因为存在更大的玩家而受到保护——无论是恶意行为者，还是普通交易者，他们的关注点根本不会落在你身上。
So if you if you were deploying something where the most amount of money that one could make if they fully exploited it was sort of in the order of low thousands of dollars, then you were sheltered by the fact that they're just much bigger fish, and and the bad actors and even the good actors or the people who are trading, etcetera, like, are just are just not you're not in their Overton window.
但随着模型变得越来越强大，而推理成本远低于顶尖安全研究人员的费用，这种长尾风险可能会迅速被暴露出来。
But as the models get better, because the cost of inference is so much lower than the cost of an extremely talented security researcher, that long tail might get shaken out very quickly.
因此，我认为风险最高的可能是那些位于长尾链上、TVL较低的小型项目，它们仍然基于像EVM和Solidity这样被充分理解的技术栈。
So so I think most at risk is probably small cap or, you know, low TVL protocols in long tail chains that are that are still built on a well understood, stacks like the EVM and Solidity.
此外，对于那些已经经过实战检验但依然非常复杂的主流DeFi合约，还存在着一种难以预知的安全风险。
And then and then I think there's there's just sort of an unknowable security risk for the for the major contracts, the the OG DeFi contracts that are currently battle tested but are still very complicated.
我们会看到，在未来一两年内，这些合约在多大程度上会暴露出来。
And and, you know, we'll see over over the coming year or two to what extent those those contracts are actually exposed.
对。
Right.
因此，那些经历了长时间市场检验、积累了大量流动性且锁定价值很高的原始合约，在短期内相对更安全，但管理这些合约的人仍需保持主动防御意识，确保在与攻击方的军备竞赛中占据上风。
So the OG contracts that have had a ton of Lindy and a ton of value locked over time that have been tested by the market are like safer in the near term, but nonetheless the people managing those contracts will still need to have agency to be on the defensive to make sure that they are winning the arms race against the offensive types.
因为攻击这些合约的收益要大得多。
Well well, the the prize is much larger for exploiting those contracts.
是的。
So Right.
我认为这会产生一种‘矿井中的金丝雀’效应——那些安全性较低但资金量庞大的协议会率先出事。
So I think it it has this, like, you know, there will probably be this canary in the coal mine effect where there will there will be smaller or or protocols that are less secure but have a lot of assets in them fall first.
而且我们需要留意第一个几乎完全由AI发起的攻击事件。
And and I think they're you know, you know, we'll have to look out for the first exploit that happens that is almost entirely from AI.
从那以后，防御方将开始全力采取必要的防护措施。
And then from there, it'll be the race will be on to start start taking the defensive measures necessary.
对。
Right.
然后，就像你说的，长尾合约的生产环境测试将不再存在，因为当攻击一个价值1000美元的合约成本只有10到50美元的代币时，这些合约根本就不会存在。
And then like the long tail, as you said, the long tail of contracts, testing in prod will no longer be a thing because when the cost to exploit a $1,000 contract is like, you know, 10 to $50 of tokens, then those contracts are simply won't exist.
有人会写一个机器人，说：‘嘿，Claude，OpenAI，去黑掉一些合约吧。’
Somebody will write a bot that says, Hey, Claude, OpenAI, go hack me some some contracts.
而这个机器人确实有能力做到这一点，因为那些以前根本没怎么认真考虑过安全问题的人——他们本来也不需要太在意——现在可就惨了。
And then that thing will actually have the capacity to do that because the people that, you know, didn't really think too hard about their security because they weren't didn't need to think too hard about their security, those people will have will not have a good day.
是的。
Yeah.
我认为这是一个普遍趋势，长尾部分将被那些善于使用AI的人收编。
I I think this is a general trend that that the long tail will get collected by
嗯。
Mhmm.
善于使用AI的人。
People who can use AI well.
比如，你可以看看预测市场，这种市场如果交易得完美无缺，你能赚到的最多也就五到一百美元。
Like, for example, you can look at something like prediction markets, where there are markets where if you trade them to perfection, the most amount of money you can make is maybe 50 to $100.
对于Jane Street这样的公司来说，派量化交易员去参与这些市场根本不划算，因为成本太高，受限于智力和注意力的成本。
And, like, it's not worth it for Jane Street to put a quant on those markets because it's too expensive for them, and it's bound by the cost of intelligence and the cost of attention.
但如果你能以仅10美分的推理成本近乎完美地交易这些市场，那你就会去做。
But if you can trade those markets near perfectly for 10¢ of inference, then you'll do it.
从整体上看，这个长尾市场可能相当有价值。
And in aggregate, maybe that long tail is pretty valuable.
所以目前，我们处于一个长尾市场因规模小而被保护的世界。
So so right now, we're in a world where the long tail is sheltered by the fact that it's small.
随着AI在所有这些领域变得越来越强大，我们应该默认所有这些都会被能熟练使用这些工具的人所整合。
And as AI gets better in all of these different domains, we should just be assuming that all of that's gonna get collected by people who are able to use these tools.
我们来谈谈EVM Bench吧。
Let's talk about EVM Bench.
这是那篇论文还是那个工具？
This is the the paper the tool?
你们管它叫工具？
You guys call it a tool?
对吗？
Is that right?
是的。
Yeah.
它既是一个基准测试，也是一个代理框架。
It's it's it's a benchmark and then also an agent harness.
所以我们可能同时发布了两个版本。
So we maybe had two releases in conjunction.
第一个评估代理利用智能合约的能力，第二个则是一个代理框架，类似于审计代理。
The first evaluates the ability of an agent to to exploit smart contracts, and then the second one is is sort of an agent harness that is, like, you know, similar to an auditing agent.
所以它实际上能够发现漏洞。
So it it can actually find the bugs.
嗯。
Mhmm.
而且，显然我们发布的这个代理工具并不具备前沿能力，因为我们不希望它被黑帽使用者利用；但我们有一个用户界面，你可以上传任何智能合约，它会进行基本的漏洞检查。
And, obviously, the agent harness that we released is sort of not at the frontier of capabilities because we don't want it to be used for for blackhats, but we have a we have a UI that that you can upload any smart contract into that will do sort of a a baseline check for bugs.
你能定义一下‘工具’吗？
Can you define harness?
我敢肯定这是个技术术语，是的。
That I I'm pretty sure that's a technical term Yeah.
我觉得程序员都懂，但我并不了解。
That I think coders will be aware of, but I'm not.
是的。
Yeah.
所以核心想法是，模型实验室会发布这些大语言模型。
So the the core idea is the model labs will release these LLMs.
对吧？
Right?
比如你会有 GPT 5.3 之类的。
So you'll have GPT 5.3, etcetera.
而且你知道，你可以进行基线测试，比如直接提示模型去问ChatGPT：嘿。
And, you know, you can do the baseline test of, like, just just prompt the models to ask ChatGPT, hey.
这个合约里有漏洞吗？
Is there a bug in this contract?
除此之外，比如说，这能让你在基准测试中达到x%的水平。
And then you in addition to that, like, you can, like, let's say that gets you to to x percent on the benchmark.
你还可以在模型周围添加一堆辅助工具，比如：嘿。
You can add a bunch of scaffolding around the model that says, hey.
举个例子，这里有一个你可以用来测试的EVM。
Like, for example, here is an EVM that you can test against.
你可以部署一个合约，实际运行一次攻击，看看是否能把钱转走。
You can deploy a contract and actually run an exploit and see if you're able to drain the money.
结果发现，如果给智能体这些工具——这种它们可以依托的辅助结构——它们的表现会好得多。
And it turns out that if you give agents these tools, this sort of scaffolding that that they can sit in, they perform much better.
所以这个工具框架就像一个基础平台，它承载着智能体和模型，并赋予它们针对该任务的特殊能力。
So the harness is like similar to basically, it holds the agent, the the model, and it gives it superpowers that are specialized to the task.
现在AI当前发展轨迹中一个有趣的现象是，我们添加的这些工具随着时间推移都会逐渐失效，因为随着模型能力提升，它会直接吸收这些辅助结构。
Now the the interesting thing on the current arc of AI is that most of these tools that we add in fall like, flake off with time because as the model gets better, it just absorbs the harness.
最典型的例子是特斯拉全自动驾驶系统初期，大部分代码都是硬编码和手工编写的，但很快就开始迅速演进，如今我认为超过50%的代码已经完全由模型驱动。
The the core example being how at the beginning of Tesla's fully self driving, talks about the majority of the code was hard coded and and handwritten and very quickly started ramping up to now, I think, over 50% of it is actually just the model.
他们已经移除了所有那些类似‘如果x，则y’的C++代码，而模型自己找到了实现目标的方法。
There's no like, they had they removed all the c plus plus code that that was, like, saying if if x, then y, and the model just figures out a way to do it.
所以目前，所谓的智能体辅助框架，还处于我们添加的‘如果x，则y’这类硬编码工具阶段，用以赋予它这些能力。
So right now, we're you know, the agent harness, quote, unquote, is in the if x, then y, like, coding hard coded tools that we're adding in to give it these capabilities.
但随着时间推移，这些功能最终会被智能体吸收，随着它不断进化。
But but probably in the fullness of time, it'll it'll get absorbed by the agent as it gets done.
明白了。
See.
所以，这个辅助框架就像一个引导程序，用来启动它，但最终数据和经验会接管机器内部的实际运作。
So, like, the harness is kinda like a bootloader to get it started, but then eventually data will take over data and experience will take over the actual internal, like, operations of the machine.
是的。
Yeah.
没错。
Exactly.
对。
Yeah.
我的意思是，目前这种辅助工具在很多看似反直觉的方面仍然非常有价值。
I mean, right now, the harness is super valuable in, like, very counterintuitive ways.
比如，事实证明，即使只是给智能体提供一个可以测试的环境，哪怕它几乎不使用，也能促使智能体思考更久、努力更甚，从而获得更好的结果。
Like, for example, it turns out that just giving an agent the ability like, an environment to test against, even if it barely uses it, leads the agent to think for longer and for it to try harder and and thus get better results.
所以，现在仍然有大量低垂的果实，因为智能体本身还没有完全参数化和校准好。
So it's like there's still so many low hanging there's still so much low hanging fruit because the agents themselves are not fully well parameterized and calibrated yet.
但当然，我们终将到达一个阶段，那时下一个版本的编解码器或Opus就能自行启动EVM，我们就不再需要这些辅助工具了。
But, yeah, I mean, definitely, we'll get to a to a point in time when, you know, the next version of codecs or or Opus will be able to just spin up an EVM on its own, and we won't need our harness for it.
好的。
Okay.
那么这个工具到底做什么？
So what does the tool actually do?
这个工具是那个负责利用或修复的智能体吗？
Is the tool the thing, like, the agent doing the exploiting or doing the patching?
还是只是用于基准测试？
Or is it just the benchmarking?
我是想了解它实际的用途。
Like, talking to me about the actual utility here.
我们最想推向世界的核心内容是这个基准测试。
The the core the core release, the core thing that we wanna get out in the world is the benchmark.
它衡量的是模型在利用智能合约方面的表现如何。
It's how good are the models exploiting smart contracts.
这个基准测试包含三个组成部分。
There are three components to the benchmark.
第一是检测漏洞的能力。
The first is the ability to detect bugs.
第二是修复漏洞的能力。
The second is the ability to patch bugs.
第三个方面，也是最具趣味性和创新性的贡献，是利用漏洞的能力。而此前在安全相关审计代理方面的尝试，最大的问题之一就是误报率过高。
And then the third, which is through the most interesting and novel contribution, is the ability to exploit bugs, which which is, you know, one of the biggest problems with previous attempts at having having security related, for example, auditing agents has been this problem around false positives.
于是代理跑来告诉你：我在合约里发现了50个漏洞，但这50个里可能只有一个是真的漏洞。
So the agent comes to you and says, I found 50 bugs in the contract, and maybe one of those 50 is an actual bug.
但你要花大量时间去逐一排查哪些是真实漏洞，这反而不如人工审计高效。
But it just it's so time intensive for you to go through and figure out which ones are real that it's not better than a human auditor.
而在我们这个基准测试的利用组件中，我们利用了加密货币可验证的特性，搭建了一个生产级的EVM环境，加载了大量链上状态，构建了一个漏洞环境，让代理尝试去利用它。
And what we did in in this sort of in the exploit component of of the benchmark is we leaned on the fact that crypto is verifiable, and we used this production grade EVM environment where we load in a bunch of chain state, and we we set up a bug environment and let the agent try to exploit it.
我们正是借助这一点，将误报率降低到了几乎为零。
We leaned on this to lower the false positive rate down to basically zero.
因此，当代理告诉你它发现了一个漏洞时，它实际上能提供一个可执行的利用证明，能在生产级EVM环境中运行，并从合约中盗取资金。
So it got to a point where if the agent tells you that it found a bug, it literally has a proof of concept that it can exploit against it can run against a production grade EVM environment and and drain money from a contract.
这正是这篇论文的核心突破：存在一个可验证的环境，能够实现极低的误报率。
And this this is sort of the core breakthrough of the of the paper is that is that there's a verifiable environment that actually leads to a very low false positive rate.
这才是真正的基准。
That's the actual benchmark.
你们已经能够有效地衡量这件事了。
It's like you guys have established, like, you guys can actually measure the thing effectively.
是的。
Yeah.
没错。
Exactly.
否则，如果有人说，‘我们发现了所有这些漏洞，在这个基准测试中得了90分’，你根本不知道这代表什么，因为确实如此。
Because otherwise, if if someone says, oh, we found all of these bugs and we got 90% on this benchmark, you don't know what it means because Right.
你无法判断其中一半是真实漏洞还是虚假报告。
You have no way of knowing if half of those are real or fake.
对吧？
Right?
因此，可验证性变得至关重要。
So the verifiability was ended up being very important.
我认为这也是模型之所以能迅速在加密领域变得极其出色的原因之一，因为你可以将与AI相关的未来大致分为两类。
I think this is one of the reasons why models are going to get extremely good at crypto very fast because the basically, you can slice the future related to AI into two categories.
一种是可验证的，另一种是不可验证的。
One is the verifiable stuff, and the other is the unverifiable stuff.
可验证的部分模型学起来非常容易，因为它们有明确的训练信号，能清楚知道自己什么时候做对了。
And the verifiable stuff is very easy for the models to learn because they have a very clear training signal, and they know exactly when they got it right.
它们可以不断重复练习，不断改进，逐步攀登这座山。
And they can just keep running at that and and improve and and climb that hill.
而不可验证的部分，就像你根本不知道自己是对是错。
Whereas unverifiable stuff, it's like you don't know there's no You don't know if you got it right or wrong.
所以，你擅长写诗吗？
So it's like, is are you good at writing a poem?
你的笑话好笑吗？
Is your joke funny?
这些方面对模型来说非常难以掌握。
Like, these are very difficult for the models to get good at.
如果你把整个代码宇宙都看一遍，找出哪个领域最可验证，你很可能最终会发现，几乎全部是加密货币领域。
And, you know, if you were to just take the whole universe of code and look at which pocket was the most verifiable, you probably would end up with pocket that's almost entirely crypto.
对吧？
Right?
整个底层都基于可验证性的概念，这意味着即使数据量很少——虽然合同形式的数据不多，加密人士在这些实验室里也极少——模型依然变得极其出色。
The whole substrate is based on the concept of being verifiable, which means that with very little data, even though we there there isn't that much in the in the form of contracts, there isn't that much in the form of, like, no crypto people are in these labs generally or very few.
由于可验证性极高，模型已经变得极其出色。
The models have gotten extremely good because it's so verifiable.
而且随着模型变得更好，比如Gemini就曾著名的仅通过上下文就学会了一整门语言。
And and also just as the models get better, like, for example, Gemini famously learned an entire language just in context.
因此，随着模型不断进步，你需要的数据量可能会变得更少。
So as the models get better, you you know, the amount of data you need might might be might be lower.
所以我认为，这些模型在加密领域的整体趋势和走向将是极其迅速地变得非常出色。
So I think the general trend and trajectory of, you know, these models are gonna get extremely good at crypto extremely fast.
我觉得我们可以确信这一点。
I think I think we can bet on that.
当EVM基准论文提到顶级模型的漏洞利用率从20%提升到超过70%时，比如最新的JATGPT Codex，这到底意味着什么？
So when the EVM bench paper says something like top models are going from 20% to over 70% exploit rate, you know, something like the newest JATGPT codex, What does that mean?
从20%到70%。
20 to 70%.
你是说70%的智能合约都会与之交互，并且会被利用？
You're like 70% of all smart contracts, it comes in contract tact with, it can exploit?
这在实际中意味着什么？
Like, what what does that mean practically?
是的。
Yeah.
实际上，这意味着我们收集了来自Code Arena等公开审计竞赛中的所有历史资金被盗关键漏洞。
Practically, it means we collected all of the historical fund draining critical bugs from open audit contests like Code Arena.
对。
Right.
所以，这些被发现的漏洞并不是那种小问题，比如有人可能让合约冻结一天之类的。
So the set of so so the bugs that are found are not like, oh, you know, you you had the small issue where where maybe, like, someone could have frozen the contract for a day or something like that.
而是更严重的问题：如果你发现了这个漏洞，就可以从这个合约中盗走资金。
It's much more you could have strained money from this contract if you found this bug.
最初20%或更少的比例意味着，如果你采用Frontier模型，并在其面前放置所有在它知识截止后出现的最困难的审计问题，它将无法发现其中绝大多数。
And the 20% initially or less than 20% initially was meant that if you took up Frontier model and you put in front of it all of the hardest all of the hardest audit problems after its knowledge cut off, it would not be able to find the vast majority of them.
到了基准测试结束时，超过70%意味着，如果你重新运行CodeArena，但不是使用GPT-4，而是使用GPT-5.3 Codex，
And by the end of the benchmark, it meant that over 70% means that, you know, if you just reran Coderena and instead of GPT four, it had GPT 5.3 codex Mhmm.
GPT-5.3 Codex将能发现人类审计员发现的70%以上的关键资金盗取漏洞。
5.3 codex would have found over 70% of the bugs, the critical fund draining bugs that human auditors found.
在我们人类审计和发现漏洞的整个历史中，它本可以发现其中70%的漏洞。
So Throughout throughout his throughout our history of human audits and finding bugs, it would have found 70% of those.
在关键漏洞方面，有一些限制条件，比如，我们并没有追溯到整个历史。
Of the critical ones, with with some constraints, like, for example, we we didn't go all the way back in history.
我们是从知识截止点之后开始的，因为我们想避免数据污染。
We started at past the knowledge cutoff of because we we wanna avoid contamination.
所以
So
是的
Yeah.
展开剩余字幕（还有 337 条）
这基本上让你得到了一个粗略的基准，也就是说，ChatGPT 5.3 Codex 大约和所有人类审计员的水平相当，达到了70%？
This kind of gets you a rough benchmark of, like, basically, chat GPT five five point three codex is, like, 70% as good as all of the human auditors out there?
差不多是这样吧？
Something like that?
差不多是这样。
Something like that.
不过，这些东西的性能是非线性的。
Although, I these things are highly nonlinear.
比如，有些愚蠢的漏洞会导致合约中的所有资金被盗，但实际上并不难发现。
Like, for example, there are dumb bugs that can lead to losing all the money in the contract, but are actually not that hard to find.
所以这就是为什么我认为论文中更值得注意的是，这些成果并不是靠模型只掌握了一种技巧就达到了70%。
So so that's why I think I think the the there there's more in the paper that is notable about the fact that these aren't there wasn't just, like, one trick that the model figured out and, like, got to 70%.
它并不是仅仅发现了重入漏洞。
It wasn't, like, all reentrancy.
对吧？
Right?
它能够发现这一系列非常多样化的漏洞。
This is a very diverse set of bugs that it was able to find.
但没错，从根本上说，这些模型已经非常接近顶尖人工审计员的水平了。
But, yes, it's like fundamentally, the the models are getting, you know, very close to being as good as the best human auditors.
拥有一个基准测试的一个有趣之处在于，似乎所有前沿实验室都热衷于竞争基准测试、赢得基准测试。
One thing that's fascinating about having a benchmark is then, like, it seems like all the frontier labs love love to compete for benchmarks, winning benchmarks.
对吧？
Right?
人文类的最后一次考试。
Humanities last exam.
事实上，他们几乎就是为了这些基准测试而进行了一些训练。
In fact, they almost, like, gained some of the training towards these benchmarks.
因此，如果你有一个有吸引力的基准测试，它甚至会在社交层面传播到所有前沿实验室。
And so if you have an attractive benchmark that kind of propagates even, like, socially to all of the frontier labs
是的。
Yeah.
那么，这似乎为它们提供了一种社会激励，促使它们进行表现和训练，以在这些基准上相互竞争、超越彼此。
Then it seems like it provides some sort of social incentive to have them perform and train in order to, you know, compete against one another to exceed each other on those benchmarks.
这是否就是你认为EVM基准所启动的飞轮效应？
Is that kind of the flywheel you feel like has been set in motion with the EVM bench?
我认为这很重要，我的意思是，从更宏观的角度来看，加密货币在其历史上一直被污名化，对AI实验室来说也难以理解。
I think I think it's I think it's an important I mean, maybe the zoomed out point is that crypto in its history has been very stigmatized and very illegible to the AI labs.
因此，迄今为止还没有出现对加密相关评估的大规模推动，这其实很荒谬，因为目前实验室完全被可验证且具有经济重要性的评估所限制。
So the fact that there hasn't already been a massive push for crypto related evaluations is kind of absurd because the labs currently today are entirely bottlenecked on evaluations that are verifiable and economically important.
而加密技术正好满足了这两点。
And crypto ticks both of those boxes.
所以我认为，是Paradigm付出了一些努力，才让这件事得以进入各大实验室的视野。
So so I think it took Paradigm pushing a little bit of our weight around to to get this through into the labs.
我认为，我们公司确实希望，这能开启一个飞轮效应，让各实验室更加关注这项技术，我们也会继续在这个领域开展工作。
And I think, yes, like, our our our firm hope is that this will start the flywheel of of labs paying more attention to this technology, and we're gonna continue doing work on this front as well.
你有什么解释吗
Do you have an explanation
为什么进展这么慢？
as why it's been so slow?
因为这也让我感到困惑，因为这些都是开源的。
Because it's also been perplexing to me because it's all open source.
所有内容都公开可用。
It's all out there.
所有数据都已经有了。
It's all like, and We already got all the data.
是的。
Yeah.
我最近和哈西德聊过，他的解释是，金融和加密领域存在很大的责任风险，尤其是当AI模型在这些数据集上进行训练时。
I mean, talked to Hassid recently and his his explanation was there's a lot of liability when it comes to finance and crypto and having AI models trained on those datasets.
对吧？
Right?
如果AI模型利用了某个漏洞怎么办？
It's like, what if an AI model does exploit a bug?
这是谁的错？
Whose fault is that?
所以也许这背后存在一些风险。
So maybe there's some risk associated with it.
嗯。
Mhmm.
你刚才提到了一种污名化现象。
There's also you you just mentioned kind of a stigma.
我觉得，是的，模因币、赌场、投机。
I think, yeah, MemeCoin, Casino, speculation.
当然，OpenAI的彼得就经历过负面事件，他将加密货币文化与许多人自称属于加密行业、试图抢先交易并开发模因币的行为联系起来，他认为这些行为很可疑。
Certainly, Peter from OpenAI, he had a negative experience that he associates with with crypto culture with much people calling themselves part of the crypto industry, try to front run him and and develop Meme Coins, and he just consider considers it shady.
那么，为什么Frontier Labs在训练这个极其丰富的数据集时如此缓慢呢？正如你所说，这个数据集完美适合训练，因为所有内容都可以被验证？
Like, what are the reasons why the Frontier Labs have been so slow to train on this incredibly rich dataset that, as you said, it's it's it's perfect for training because all of it can be verified?
我的感觉是，这几乎完全是社会因素造成的。
My sense is that it's almost entirely a social thing.
我的意思是，在我的朋友圈里，加密货币是迄今为止最具反主流趋势的行业。
I mean, in in my peer group, crypto is the biggest industry that has remained the most contrarian.
我认为部分原因在于它的声誉波动性很强，另一部分原因在于，这个行业里顶尖人物与普通人的差距，比其他任何行业都要大得多。
And I think part of that is because it's very reputationally volatile, and part of it is because there's this dynamic where the best people in the industry the gap between the best people in the industry and the median person in the industry is much larger than anywhere else.
所以，如果你没有接触到加密货币中高质量的那一小部分，那你看到的就全是骗局，这会扭曲你的看法，让你完全否定整个行业。
So if you, for example, don't have exposure to the high quality pocket of crypto, then all you see are the scams, and it's it's like that can distort your view such that you you just completely dismiss the industry.
历史上，这里面一直存在大量的超额收益。
And historically, there's been a lot of alpha in that.
对吧？
Right?
我认为我们很多人都从这种显著的声誉波动中获益了。
And I think a lot of us have benefited from the fact that that there's significant reputational reputational volatility.
如果你这个人对声誉波动不那么敏感，性格上也不太在意这些，那你就能在加密货币领域做得非常好。
And if you aren't as sensitive to that as a person in terms of your temperament, you can do very well in crypto.
但我认为这本质上是一个社会问题，关键在于至今还没有一个品牌能够成功连接加密货币和人工智能这两个世界。
But I think I think it's a social thing, and I think it's just this legibility point about about there just hasn't been a a brand that can bridge the crypto and the AI worlds.
我觉得，只要和加密货币沾上边，在AI领域历史上就会被玷污。
I think it's like if something touches crypto, it it sort of in the AI world historically has been tarnished.
因此，人们都试图完全避开它。
And as a result, people just have tried to try to avoid it altogether.
这反而为像EVM Bench这样的项目创造了机会，让它能在OpenAI内部被构建和发布，而我认为，所有主要的模型实验室都将使用这个基准，未来任何版本也大概率会如此，且实验室内部几乎没有竞争。
And this is sort of this has created an opening for something like, for example, something like EVM Bench to get built and shipped inside OpenAI without any sort of I think, like, all of the major model labs are gonna be running on this benchmark and probably any future versions of it without significant competition inside the labs.
实际上，并没有五六十个与加密货币相关的基准或训练环境被广泛采用。
Like, there aren't, like, 50 30 to 50 crypto related benchmarks or or training environments that people are shipping.
从某种意义上说，这反而增强了我们的主动性，因为他们会直接依赖加密货币行业来判断什么对他们有价值。
In some sense, it's actually agency inducing for us because because they'll just defer to to the to the crypto industry to just figure out what's valuable for them.
但我认为，这本质上是个社会问题，与这些动态密切相关：比如，你看到某人变得极其富有，而你认为他根本不该这么富有。
But I think it's fundamentally a social issue, and it's it's sort of tied to all of these dynamics around, like, you know, you see someone who gets extremely wealthy who you don't think should get extremely wealthy.
比如，这个行业波动性很大，可能有个你并不尊重的人赚了很多钱。
Like, maybe it was like there's a lot of volatility in the industry and, like, there's some person who you don't respect who you who made a lot of money.
这些种种想法，正是这些实验室里的AI研究人员心中所想，导致他们认为整个行业都是骗局。
Like, these are all kinds of things that that go on in the minds of, like, the AI researchers at these labs that lead them to think that the whole industry is a scam.
而且，你知道，这使得投资加密货币成为了一个绝佳的环境，因为硅谷的任何人都没有对此开放。
And, you know, obviously, there's this has made it an incredible environment to be investing in crypto because it's just not in the open window of anyone in the valley.
但我认为这是一个核心的动态。
But but I think I think that's a core dynamic.
有意思。
Interesting.
有意思。
Interesting.
因为我们知道，机器人、AI大语言模型非常擅长编写代码。
Because we know bots, AI LLMs, they are very good at writing code.
这几乎是它们最先掌握的能力。
It's like the first thing that they got good at.
这一点在编写EVM代码时也成立吗？
Is that the same is it also true with writing EVM code?
在编写全球其他代码与专门编写EVM代码之间，是否存在差距？
Is there a gap there between writing the rest of the world's code and EVM specifically?
是的
Yeah.
我的意思是，历史上确实如此。
I mean, historically, there has been.
而促使我们开展这项工作的一个原因，是意识到这些模型在Python方面非常出色，但在Solidity上却很糟糕，对像Solana相关的代码也表现得很差。
And part of the reason, you know, one component of what motivated us to start this work was the realization, like, man, these models are so good at Python and so bad at Solidity and so bad at, for example, like, Solana related code.
老实说，任何涉及加密货币的东西都一样。
Honestly, anything that touches crypto.
当时我的预期是我们不得不从行业里收集大量数据，然后一点点喂给实验室，让模型真正掌握这些内容。
And part of my expectation at the time was that we were gonna have to go, like, pull crowdsource a bunch of data from the industry and, like, spoon feed it into the labs to get them the models really good at this.
但结果发现，由于区块链代码的可验证性很强，再加上这些模型本身具有通用性，它们在远少于预期的输入数据下，进步速度远超我们的预料。
But it turned out that because the substrate is so verifiable, and also because there's sort of generality in these models, they ended up getting quite good much faster than we expected with much less input than than we expected.
所以存在这样一种现象：如果你教一个模型用英文写诗，再教它用西班牙语学生物学，它就能学会用西班牙语写诗，即使你从未向它解释过如何用西班牙语写诗。
So there's this dynamic where if you teach a model a poem in English and then, you know, biology in Spanish, it figures out how to write a poem in Spanish even though you never describe how to write poetry in Spanish to the model.
我认为这里也发生了类似的动态——模型在没有大量直接训练数据的情况下，某种程度上‘学会了’加密货币的语言。
And I think that kind of dynamic is happening here as well where where it's sort of like, quote unquote, learning the language of crypto without without as much direct training data.
而且，我也觉得很难低估这件事的可验证性。
And also just because I like, very hard to, like, under underrate the verifiability of the thing.
对吧？
Right?
大多数软件都很难验证。
Most software is is hard to verify.
你需要人工标注者去检查：这个东西是否正确？
You need human label labelers to go in and check, is this thing correct?
它在运行吗？
Is it running?
你所能依赖的可验证性门槛，往往只是程序能否编译通过，以及能否通过测试。
Kind of the only threshold of verifiability you have is does the program compile, and does it pass the tests?
但这些测试本身还是需要由人工来编写。
But the tests need to be sort of written by a human.
对吧？
Right?
你并没有这种概念，比如我们有一系列状态。
You don't have this notion of, like, we have a bunch of state.
我们可以对这些状态做出断言。
We can make assertions about the state.
例如，我们可以将一个模型部署到一个全新的、从未见过的EVM合约上，并断言它是否能够‘提取’资金。
We can, for example, send a model on a new contract, on a new EVM that's never seen before and make assertions about whether it's able to, quote, unquote, drain money from it.
这些概念原本都需要被硬编码到程序中。
Like, those concepts all would have otherwise needed to be hard coded into the program.
但因为这是加密货币，而且有这么多标准。
But because it's crypto, we and there's so many standards.
它是可验证的，而且模型的能力正在迅速提升。
It's verifiable, and the models are just ramping up really quickly in capabilities.
好的。
Okay.
所以你觉得，现在正是加密货币数据开始大量涌入、用于训练这些模型的时机吗？
So do you think that this right now is the time that the the the floodgates of crypto data to train these models that that is now starting to open and going to be opened?
如果是这样，你期望未来模型会具备哪些能力？
And if so, what types of capabilities do you expect future models to kind of drop?
我的意思是，是的。
I mean Yeah.
它们会不会在一些核心大语言模型中直接集成针对加密货币的技能和角色连接器？
Will they have skills and, I guess, personas connectors directly for for crypto within some of the the core LLMs?
或者，你期待哪些类型的发展？
Or what types of developments are you looking forward to?
我们之所以从安全性开始，而不是通用编程能力，是有原因的。
There's a reason we started with security and not sort of general programming capabilities.
这是因为加密安全具有非常明显的经济价值。
It's because it has this very nice shape of it's extremely economically valuable.
它也非常依赖智能水平。
It's extremely sort of intelligence bound.
对吧？
Right?
就像你无法做到那样。
It's like you can't yeah.
这是智力密集型的，并且非常容易验证。
It's intelligence bound, and it's very easily verifiable.
所以我们知道漏洞何时发生。
So we know when an exploit has happened.
因此，安全能力预计会迅速发展，而我们已经讨论过这所有的影响。
So security capabilities, expect, will develop very quickly, and and then we've talked about all the implications of that.
其他与加密相关的功能，我认为，比如机制设计领域或市场相关的问题，例如交易所的机制是什么？
Other crypto related capabilities, I think, for example, things in the domain of mechanism design or around market market related films, like, what is the mechanism for an exchange?
如果你有一群智能体组成的市场，那么它们之间最合适的协调方式是什么？
How how do if you have market of agents, right, what is the best way through which they should coordinate with each other?
这些我认为是充满潜力的领域。
These are, I think, open fertile soil.
当然，你还可以深入到协议层。
And then, of course, you can go down to the protocol layer.
对吧？
Right?
你可以说，模型是如何将一笔交易写入以太坊区块链的？
You can say, well, how does a model land a transaction on the Ethereum blockchain?
在以太坊客户端或协议客户端层面存在安全问题，没错，也许有价值一千亿美元的资产存放在开源智能合约中，但以太坊和Solana的市值远超这个数字，如果你能发现Geth、Rust等关键漏洞，就能利用这些漏洞进行攻击。
How like, there's a security side at the at the Ethereum client level, right, or at the protocol client level where, like, sure, maybe there are a $100,000,000,000 of assets sitting in open source smart contract, but there's way more than that in ETH and SOL market cap that can be exploited if you're able to find critical vulnerabilities in GATH, RATH, etcetera.
所以我认为深入协议层将会非常重要。
So so I think going down into protocol layer is gonna be important.
模型在MEV以及各种套利策略方面的能力，我认为会像长尾攻击一样产生同样的影响。
Model capabilities around MEV and and sort of extractive extractive tactics, I think that will have the same effect as, you know, long tail hacks.
对吧？
Right?
链上有很多东西，只要你能够完成端到端的流程——发现市场中的Alpha、构建交易、承销交易，然后可靠地提交并上链交易，你就能轻松获取这些收益。
There's a bunch of stuff on chain that you can just collect if you if you're able to do the end to end process of figure out alpha in the market, construct a trade and underwrite it, and then submit the transaction and land it on chain reliably.
这些能力目前模型还不是很擅长，但它们会很快变得非常出色。
These are all things that actually the models are not that good at right now, but they will get good at really quickly.
所有这些看起来对加密货币的长期发展都是有利的。
And all of that does seem long term good for crypto.
然而，就当下和短期而言，比如当我们读到EVM Bench论文中提到顶级模型的漏洞利用率已从...提升到70%以上时。
However, the the here and the now and and the short term, like when we read in the EVM EVM Bench paper that top models are going from to over 70% exploit rate.
我不确定是该为此感到高兴还是担忧，因为嗯。
I'm not sure whether to, like, I mean, feel good about that or bad about that because Mhmm.
这在一定程度上取决于意图以及谁在利用它。
It it sort of depends what the intent is and who is harnessing it.
如果是白帽，那就太好了。
So if it's whitehat, that's great.
这意味着我们能在攻击者之前更早地发现漏洞和利用方式。
I mean, that improves our ability to find bugs and exploits before attackers do.
但如果是黑帽，那就不太好了，因为这会提升他们比我们更早发现这些漏洞的能力。
However, if it's blackhat, that's not great because, you know, it improves their ability to find these before we do.
所以嗯。
So Mhmm.
这取决于，我认为，对于这套工具来说，关键在于谁在使用这些工具以及他们的意图，这决定了它是短期到中期的看跌还是看涨。
It depends, I guess, with with this toolset, it depends who is using the tools and the intent behind them as to whether it is short to medium term, bearish, or bullish.
我想，它确实给市场带来了一些波动性和不确定性。
I guess one thing it does is it does seem to inject some variance and uncertainty into the market.
我现在在人工智能领域处处都能感受到这种感觉。
I'm feeling that everywhere with AI right now.
它可能非常好。
It's just like, it could be really good.
也可能有点糟糕。
It could be kind of bad.
但有一件事是肯定的，那就是它不会再那么无聊了。
But one thing, it's it's not going to be as boring.
结果将会是高度多变的。
It's going to be highly variant outcomes.
当你想到这里所释放的能力，以及大语言模型检测漏洞的能力时，这对加密货币在短期到中期是好是坏呢？
I guess when you think about the capability that is being unlocked here and the ability for LLMs to detect exploits, is that good or is that bad for crypto in the short to medium term?
是的
Yeah.
我的意思是，你提到在长期来看，加密货币几乎对所有这些发展都具有正向杠杆效应。
Well, I mean, you mentioned in long term, crypto's positively levered to almost all of these developments.
随着模型在安全性方面变得极其出色，这将提升整个行业的上限，因为
And as models get extremely good at security, this will raise the ceiling for the whole industry because
而且这是因为这将变成一种适者生存的局面。
And and that's because it's, like, gonna be a survival of the fittest thing.
对吧？
Right?
因为所有薄弱的东西都会被淘汰掉
Because all the weak stuff gets just
所以这取决于我们。
So I it's up to us.
我认为这取决于我们，也就是整个行业，决定通往那条道路的方向。
I think it's up to us, like, what the path there is, us being the industry.
这可能是适者生存。
It's it may be survival of the fittest.
也许我们能找到一种方法，让防御措施领先一步。
It may be that that we get figure out a way to have the defense get ahead.
但我认为核心观点是，这些网络上能够维持的资产数量与它们的安全性成正比。
And I think but the core point is that the amount of assets that can be sustained on these networks is proportional to how secure they are.
从长远来看，有一个好处：随着安全性提升，更多资产能够安全地留在链上。
And in the long term, there's this benefit where as security improves, more assets will be able to securely stay on chain.
但在短期内，我认为这是一件掌握在我们手中的事情。
Now in the short term, I think I think this is one of those things where it's in our hands.
对吧？
Right?
我认为，如何最好地处理这个问题，取决于行业的主动作为。
It I think it's bound by the industry's agency on the best way to handle this.
我们不知道，如果只是让时间自然推移，会存在很多不确定性。
We don't know like, if we just let the clock play forward, there's a lot of uncertainty.
我们不知道攻击者，也就是黑帽黑客，是否会比白帽黑客更早获得这些能力。
We don't know exactly who whether the attackers, you know, the blackhats will get capabilities before the whitehats do.
但我们也积极参与这个市场，我们可以推动这一趋势，例如确保如果出现前沿模型、未发布模型，或与AI相关的安全新进展，我们能将这些成果融入顶级协议中。
But we also are active participants in this market, and we can we can bend the arc of this such that, for example, we make sure that if there are frontier models or unreleased models or there are there are new developments in security relating to AI that we that we get this into the top protocols.
你可以想象的一种短期到中期的情景是，每个智能合约都会被攻击方和防御方全天候扫描。
You know, one one version of the world that you can imagine, the short to medium term, is that you always have every single contract being scanned by both adversarial actors and defensive actors twenty four seven.
一旦发现漏洞，谁先捕捉到，谁就会相应地采取行动。
And when there's a bug that surfaced, you know, whoever catches it first sort of will will react accordingly.
在这样的世界里，这纯粹就是好人和坏人之间的竞赛。
And then in that world, it is just kind of more of a race between between the good guys and the bad guys.
我认为我们在确保好人在这场竞赛中领先方面，拥有非常有利的条件。
And and I think we have a lot of a pretty great hand in terms of in terms of making sure that the good guys have have the lead in that race.
到最后，当所有薄弱的合约都被利用，或者我们的安全措施已经足够强大，使它们无法被利用时。
At the end of this, once all of the contract, like the weak contracts have been exploited or we've beefed up security enough such that they're not exploitable.
我想这将为我们世界带来一个极其坚固的金融系统，一个超安全的系统。
I guess that gives us an incredibly hardened financial system for the world, something that's ultra security.
这几乎可以说是近乎完美了。
There's almost like a it's close to perfect.
对吧？
Right?
这到底能达到几个九呢？
It's like how many how many nines.
我的意思是，可能有四个九，甚至五个九。
I mean, it's maybe like four nines, five nines.
这几乎创造了一种针对全球金融资产的安全性双峰模型。
And it almost creates kind of a a barbell model of a security for for the world for financial assets.
最安全的金融资产很可能就存在于这种黑暗森林环境中，也就是链上。
It's like the most secure financial assets will probably be in this dark forest environment, like on chain.
我们怎么知道这一点？
How do we know that?
因为全世界已经倾尽所有力量攻击它，包括我们最智能的LLM，但它依然屹立不倒。
Because the world has thrown everything it can at it, including our most intelligent LLMs, and it's still there.
它依然屹立不倒。
It's still standing.
对吧？
Right?
从未被攻破过。
Hasn't been exploited.
所以这将是杠铃的一端。
So that'll be one side of the barbell.
另一端，老实说，是完全脱离数字世界的东西，比如一块金条之类的实物资产。
The other side, honestly, is things that are completely outside of the digital world altogether, you know, like a clump of like a a bar of gold or something that
数字黄金和实物黄金。
Digital gold and actual gold.
对。
Right.
然后中间的所有东西
And then everything in
中间部分会很容易被利用，非常不安全。
the middle will be pretty exploitable, pretty insecure.
我不禁想，这是否就是我们世界未来的发展方向。
I wonder if that's what is on the horizon with with our world.
是的。
Yeah.
直到大语言模型学会如何合成黄金。
Until the LMs figure out how to synthesize gold.
对吧？
Right?
没错。
It's Right.
对吧？
Right.
对吧？
Right.
然后派机器人去寻找黄金。
And send the robots after the gold.
对。
Right.
是的。
Yeah.
我想，我觉得大概这样说得通，我认为这种世界的双峰观点可能会成为现实。
I think I think that I guess that makes sense, and I think that that barbell view of the world might be how things play out.
我想，我看待加密货币这个行业和技术的方式是，如果你从第一性原理的角度出发，比如说，你想以光速进行支付。
I guess the way that I relate to crypto as an industry and also as a technology is is that if you start from the first principles vantage point of let's say you wanna do payments at the speed of light.
对吧？
Right?
比如，我想从美国给你，瑞安，转账到欧洲或其他地方，就像发邮件一样快。
Like, I wanna send you, Ryan, money from America to Europe or some other part of the world, and I send it as fast as I send an email.
而你面临的问题是，你不知道我是否也把这笔钱转给了大卫。
And the problem that you have there is that you don't know if I also sent that money to David.
对吧？
Right?
你遇到了双重支付问题。
You have this double spend problem.
而比特币正是解决了这个问题，将交易时间缩短到了大约一小时。
And this was what Bitcoin solved, and it got the time for that transaction down to about an hour.
从那以后，我们不断取得进展，不仅提升了这些交易的速度，也增强了它们的表达能力。
And since then, we have had successive developments that have increased both the the speed of these transactions and the expressivity of these transactions.
我认为，在这种世界观下，加密行业并非偶然发展成今天的样子，而是如果你从第一性原理出发，推演未来，它必然如此。
And I think that I think that in that worldview of this is like this is not just some path dependent thing that happened that the crypto industry emerged the way it did, And that actually this is if you were to play it forward from first principles that this is how it has to be.
你会得出这样的结论：如果代理方希望以互联网的速度进行移动，而当前的银行系统是在汽车发明之前建立的，那么这些代理方最终会发现加密网络才是进行交易的正确方式。
You end up in this in this conclusion where, you know, if you have agents that wanna move at the speed of the Internet and the current banking system was, you know, created before cars were invented, that that these that those agents are going to discover the crypto rails as as the right way to transact.
而且，我想大概六个月到八个月前，至少对我而言，还不清楚代理方是否能足够熟练地掌握与加密相关的软件，从而发现现有的网络，也许他们不得不从头开始重新发明。
And, you know, I think there was a concern maybe, like, six to eight months ago, definitely on for me, where it was not clear if the agents would get good enough at crypto related software, for example, for them to be able to discover the current rails, and maybe they'd have to reinvent them from scratch.
但在过去六个月到八个月里，随着我们与OpenAI合作推进这项工作，越来越清晰的是：第一，加密领域存在极其强大的网络效应；第二，这些代理方能够迅速学习——他们就是想学习这些可验证的东西，而加密正位于这一列表的顶端。
But over the last six to eight months, as we've been working on this work with OpenAI, it's become increasingly clear that that, one, there are extremely strong network network effects inside of crypto, and that and two, these agents are able to just learn like, they just want to learn these verifiable things, and crypto is very high on that list.
所以到目前为止，我对加密货币作为这些智能体的底层平台变得极其非常乐观，我认为这还是一个开放的游戏。
So so at this point, I think, you know, I've become extremely, extremely bullish on on crypto as a substrate for for these agents, and I think it's sort of an open game.
那么在加密领域，谁会胜出呢？
Who in crypto is going to win that?
但这项技术的形态完美契合了这一点。
But but the shape of the technology fits it perfectly.
EVM 是加密领域中最常见的编程语言和编程环境，Solidity 是最常用的语言。
The EVM is by far the most common programming language, programming environment in in crypto, Solidity being the most common language.
还有一些小众语言，比如 Cardano 使用的是 Haskell。
And then there's like some long tail languages like Cardano is like Haskell, for example.
正如我们所知，人工智能喜欢数据，数据越多，AI 就能表现得越好。
And as we know, AI loves data, the more data and AI can get on the better better it can be.
你如何看待围绕这些环境的网络效应？
What do you think this network affects around environments?
这对人工智能意味着什么？
How does that play into AI?
比如，EVM 会成为 AI 最青睐的运行环境吗？
Like, is the EVM going to be the favorite environment for AIs to work in?
Solana 的那个，Solana 到底是什么？
Does Solana's what's what's Solana?
SVM。
SVM.
SVM。
SVM.
SVM。
SVM.
Solana 的 SVM 也会跨越那个门槛吗？也许‘门槛’这个说法本身是个错误的比喻？
Does the Solana's SVM also, you know, cross the threshold maybe the threshold thing is a is a false illustration?
你怎么看？
What do you think?
我觉得真的很难说，目前局势还很不明朗。
I think it's it's actually very hard to say, and currently, the ball is in there.
我们不知道。
We don't know.
部分原因在于，我在开发EVM Bench时遇到的一些瓶颈，当时我们其实已经开始着手Solana相关组件的工作。
Part of the reason why well, I mean, I can point to some of the bottlenecks we had while we were developing EVM Bench, where we also have actually started work on on the Solana related component of it.
但举个例子，一个挑战是，实际上需要大量的人才来构建这些评估体系。
But for example, one challenge was that actually it it requires a lot of human talent, like human talent to be able to go in and and construct these evals.
而对我们来说，找到熟悉Solidity的人要容易得多。
And that was just much easier for us to come by for solidity.
这些工作真的很难做。
And these things are really hard to build.
对吧？
Right?
这些都是相当复杂的基础设施。
These are pretty heavy infrastructure.
所以，即使是很小的摩擦因素，也会导致你在需要缩减范围时，不得不优先放弃那些更难实现的部分。
So so even small additions of friction lead to, you know, when you need to cut scope in some capacity that you kind of have to cut it in in the direction of the stuff that you can do more easily first.
话虽如此，我认为关键在于两点：第一，加密货币的可验证性是一个巨大优势；第二，这些模型在学习新编程语言时对数据的需求正变得越来越少。
That being said, I think the point about the fact that, you know, one, crypto's verifiability is a huge edge, and two, these models are becoming less and less data hungry when it comes to learning new programming languages.
我认为，最终的结果可能会比表面上看起来更加公平。
I think it may end up being more even of a playing field than it might seem.
而且，至少我们有明确的意图和兴趣，要跨越不同生态系统，深入到协议层，打造一个更全面的加密货币旗舰基准测试。
And I think, you know, at least we have have an intent and interest in actually going across ecosystems and down the stack to the protocol layer and and making this sort of more expansive of a crypto flagship benchmark.
但没错，目前EVM周围显然存在网络效应。
But, yeah, right now, there are obviously network effects around the EVM.
对。
Right.
另一个反直觉的例子是，为什么像Solana这样的平台有可能迎头赶上：乍一看，那里大多数合约都是闭源的，这似乎非常不利。
One other example of a counterintuitive reason why why someone like Solana might be able to catch up is that first at first glance, it may seem really bad that most of the contracts there are closed source.
但如果你从另一个角度看待：如果代码是开源的，就会被纳入训练集，那么闭源的基准、训练数据和合约反而可能对模型发展更有价值，因为它们目前不在训练数据中——比如，我们在评估的各个部分都设置了所谓的‘金丝雀标签’，用来将这些内容从大多数模型的训练过程中过滤掉。
But if you take the worldview of actually, if it's open source, it gets in the training set, then the closed source benchmarks and training and contracts for training might actually be more valuable for for a model's development because it's not in the right now, for example, like, we have these sort of what are called canary tags in various parts of our evaluation that filter them out of the training process of most models.
所以，你知道，你可以用一些技巧，但EVM基准中的漏洞随着时间推移，仍有可能渗入到模型的预训练阶段。
So, you know, there are tricks you can do, but still it's possible that these the bugs in EVM bench over time leak into the pretraining of the models.
而如果是闭源的，就根本不会泄露进去。
Whereas if it were closed source, it would not leak in at all.
所以我的预期是，起初可能会有一些不对称性，但实际上模型会非常擅长所有这些方面。
So so my expectation is actually that, you know, there will be some asymmetry at first, but actually the models will get really good at all of it.
而且会有一种基于实力的机制，让真正优秀的人脱颖而出。
And and there will be sort of a merit based sort of right who, you know, who the best will rise to the top.
在获利后，你是否持有USDT或稳定币，却不知下一步该如何配置？
Are you sitting on USDT or stablecoins after taking profits and wondering where to deploy next?
如果你能在不离开加密货币的情况下投资股票、黄金和ETF，会怎样？
What if you could access stocks, gold and ETFs without ever leaving crypto?
这就是BITGET上代币化股票所实现的功能。
That's what tokenized stocks on BITGET unlock.
传统市场仍然全天候运行，但资金正24/7地迁移到链上。
Traditional markets still run unlimited hours, but capital is moving on chain 20 fourseven.
在BITGET上，你可以24/7交易代币化股票和ETF，最高可达100倍杠杆，所有交易均直接以USDT或USDC结算。
On BITGET, you can trade tokenized stocks and ETFs 20 fourseven with up to 100x leverage, all settled directly in USDT or USDC.
无需经纪账户，无需提币，无需切换平台。
No brokerage accounts, no off ramps, no platform switching.
BITGET已处理超过180亿美元的代币化股票交易量，其中大部分发生在过去一个月内。
BITGET has already processed over $18,000,000,000 in tokenized stock trading volume with most of that happening in the past month alone.
该平台目前占据了Ondo代币化股票现货市场近90%的份额。
The platform now captures close to 90% of Ondo's tokenized stock spot market share.
随着黄金和白银创下历史新高，链上交易也随之增长。
As gold and silver hit record highs, on chain trading followed.
在过去两周，与白银挂钩的SVL和与黄金挂钩的IAU在BITGET上的交易量激增。
Over the past two weeks, volume in SVL on tied to silver and IAU on linked to gold surged on BITGET.
这正是BITGET通用交易所愿景的体现。
This is BITGET's universal exchange vision in action.
将加密资产与现实世界资产融为一体，依托原生加密的速度与灵活性构建。
Crypto equities and real world assets in one place built with crypto native speed and flexibility.
如果你想像交易加密货币一样交易股票，不妨在BITGET上探索代币化权益产品。
If you want to trade stocks the way you trade crypto, explore tokenized equities on BITGET.
点击节目笔记中的链接以了解更多信息。
Learn more by clicking the links in the show notes.
这不是投资建议。
This is not investment advice.
在加密领域，很少有人在公开预测顶部或底部时真正押上真金白银。
Few people in crypto put real skin in the game when they make public top or bottom calls.
DeFi Report 就是其中之一。
The DeFi report is one of them.
在10月10日闪崩前一周，DeFi Report 的迈克尔通过邮件向整个通讯订阅者表示，他将大幅降低风险，将大部分持仓从加密货币转为现金。
The week before the October 10 flash crash, Michael from the DeFi report emailed his entire newsletter saying he's going aggressively risk off and sold the majority of his book from crypto into cash.
当时以太坊价格约为4000美元，比特币约为11000美元。
This is when ETH was about $4,000 and Bitcoin was a 110.
迈克尔运营着DeFi Report，这是一个以数据、周期洞察、风险管理、透明度，最重要的是——真金白银的投入为基础的行业领先研究平台。
Michael runs the DeFi report, an industry leading research platform built on data, cycle awareness, risk management, transparency, and most importantly, skin in the game.
我们在Bankless很喜欢迈克尔。
We like Michael at Bankless.
我们喜欢他的分析，这就是为什么你大约每月都能在Bankless播客中听到他。
We like his analysis, and that's why you hear him on the Bankless podcast about once a month.
DeFi报告正在为Bankless的听众提供一个月的免费访问权限。
And the DeFi report is giving bankless listeners one free month of access to the DeFi report.
所以，如果你正在寻找一些敏锐的、以数据为驱动的分析，以便对你的投资组合做出更明智的决策，你可以在DeFi Report Pro中了解Michael是如何预测顶部的，以及他接下来的计划。
So if you're looking for some sharp data driven analysis to make better informed decisions around your portfolio, you can learn why and how Michael called the top and what he's doing next all in the DeFi report pro.
去了解一下吧。
Check it out.
链接在节目说明中。
There is a link in the show notes.
以太坊的目标是在未来彻底验证其整个技术栈。
Ethereum's aspiration is to formally verify its entire Mhmm.
在适当的时候，实现端到端的技术栈验证。
End to end tech stack in the fullness of time.
你知道的。
You know?
首先，我们需要先获得像beam chain这样的东西。
We need first, we have to get, like, the beam chain.
我们必须付出巨大的努力才能达到那里，但最终，我们希望对整个以太坊技术栈进行形式化验证。
We have to do all all the the hard force to get there, but ultimately, we want to do a formal verification of the entire Ethereum tech stack.
嗯。
Mhmm.
基于人工智能的形式化验证。
AI based formal verification.
这真的存在吗？
Is that is that a real thing?
人工智能的能力是如何融入形式化验证的讨论中的？
How does AI capabilities work its way into the conversation of formal verification?
是的。
Yeah.
我的意思是，我认为这确实是一个真实存在的东西。
Well, I mean, I think it is I think it is a real thing.
去年，我们投资了一家名为Harmonic的公司，该公司由Vlad Tavov、Robin Hood和Tudor共同创立，专注于数学模型。
Last year, we invested in a company called Harmonic, which is a foundation of math model cofounded by Vlad Tavov, Robin Hood, and Tudor.
我认为他们的部分理念，以及世界明显的发展趋势是，现在生成的软件数量已经远远超出了人类能够审查的范围。
I think part of their thesis and I think part of where the world is clearly going is that is that there's more software that is being generated than can be possibly reviewed by humans.
形式化验证是快速检查软件组件是否真正实现其声称功能的一种方式。
And formal verification is one way to quickly check whether a component of software is actually doing what it says it's doing.
当然，在安全背景下，如果规格书编写得当，形式化验证可以带来质的飞跃。
And then obviously in the context context of security, it it can especially if the spec is written correctly, it can be it can be a step function change.
但它并不是万能的，因为你仍然需要为形式化验证编写规格说明。
Now it's not a silver bullet in the sense that you still have to write the spec for the formal verification.
因此，仍然存在漏洞渗入的空间。
So there's still surface for bugs to get in there.
但你可以论证，编写形式化验证规格说明的漏洞面，可能比直接编写代码的漏洞面更小。
But, you know, you can make the case that actually the surface for bugs in writing a formal verification spec might be lower than writing the code to start with.
而且随着时间推移，我认为所有顶尖的模型和软件最终都会被形式化验证。
And and definitely with time, I think all of the best all the best models all of the best software will probably end up being formally verified.
如果你从智能体的角度来看，面对两个选择，其中一个经过了形式化验证，另一个没有。
And and if you take the vantage point of an agent and you have two options to choose from, one of them is formally verified and one of them is not.
而经过形式化验证的那个可能会仅仅因为具备这些优良特性而获得优先选择。
And the formally verified one might just gain preference just because it has all these nice properties.
在EVM基准测试论文中，有一节叫做‘未来方向’。
There's a section in the paper of the EVM bench paper called future directions.
EVM基准测试v2会是什么样子？
What does EVM bench v two look like?
你称这个为一个项目，它具体是指什么？
How does this you've called a project?
这个项目接下来将如何发展？
How does this project grow from here?
我们的最高目标是帮助模型实验室提升其模型的加密能力。
The top level goal that we have is to help the model labs develop the crypto capabilities of their models.
我认为安全性是其中的一个组成部分，而且可能是越来越紧迫的一个部分。
And I think that security is one component of that and maybe an increasingly urgent component of that.
但EVM Bench还有很多没有涉及的地方。
But there there's so much that EVM Bench does not touch.
因此还存在其他生态系统和堆栈。
So there are other ecosystems and stacks.
有协议层，我们之前讨论过，从安全角度来看，以太坊协议的安全性可能比任何特定的Solidity合约都更为重要。
There is the protocol layer, which we talked about, where maybe arguably it's more important from a security standpoint that the Ethereum protocol is secure rather than any specific solidity contract.
还有协议外的组件，比如如何将交易上链？
There there are out of protocol components, like how do you land a transaction on chain?
如何处理内存池？
How do you deal with the mempool?
如何应对加密货币中那些非确定性的部分？
How do you deal with sort of the nondeterministic parts of crypto?
当然，还有更远离可验证性和智能边界的一些组件，比如密码学和零知识证明等。
And and then, obviously, there's there are components that are even farther on the verifiable and intelligence bound trajectory, like, for example, around cryptography and around zero knowledge proofs, etcetera.
我认为所有这些领域都是未来研究的绝佳土壤。
And I think all of these are extremely fertile soil for future work.
所以我们目前正在积极寻找合作者，共同开发 EVM Batch 的未来版本，当然我们自己也在推进下一步工作。
So we're currently open to trying to source collaborators for for future versions of of EVM Batch, and and, you know, we're obviously working on on next steps for it ourselves.
我认为，我们终于在模型实验室中站稳了脚跟，成功将加密货币能力引入前沿模型。
And I think that this this direction of, like, we finally have a foot in the door into the model labs for for getting crypto capabilities into the frontier models.
我认为整个行业都应该利用这一契机，尽最大努力提升这些模型在加密领域的表现。
And I think that we should we should leverage that as an industry, and we should try to try to get these models as good as at crypto as we possibly can.
阿尔平，你非常聪明，这一点从你的论文和今天的对话中就能看出来。
Alpin, you're you're very smart as evidenced by this paper in in this conversation.
你显然具备很强的自主性和行动力。
You have a high degree of of agency, obviously.
你毫无疑问站在了前沿。
You're definitely on the frontier.
你选择留在加密领域，而不是离开去投身人工智能。
You've chosen to stay in crypto and not kind of leave and go to AI.
你对加密领域表现出极大的乐观，甚至认为在当前这个时刻，这种乐观本身就是一种反主流观点。
And you seem incredibly bullish at crypto, even bullish that it's that it's contrarian at this moment in time.
对你个人而言，为什么选择加密货币？
Why crypto for you personally?
我个人从未在心中对行业设过明确的界限。
I've personally never had hard lines around industries in my mind.
我觉得，你知道，我们
I think, you know, we
我们谈论‘我做X是因为’，只是为了让别人更容易理解我们的行为，但我不认为这是与之建立关系的正确方式。
we talk about, like, I work in x because like, to make what we're doing legible to other people, But I don't think that's the right way to relate to it.
我之所以一直投身加密领域，一方面是因为它极其富有智力上的吸引力，另一方面正如我所说，它在我的那些最聪明的朋友中依然保持着极强的反主流立场，而且我能明确指出他们错在哪里。
I've spent all all of this time in crypto because it's been, one, it's been extremely intellectually interesting, and two, it has this it's just as I mentioned, it's remained extremely contrarian among my smartest friends in a way in ways where I can put my finger on exactly what they're missing.
我认为，这大概就是一个人所能追求的最好状态了。
I think that's sort of that's kind of the best that that one can ask for.
我觉得，我们之前谈到加密货币如何受益于人工智能的安全发展，但事实上，你也可以论证，它目前正受益于世界上绝大多数的发展趋势。
I think that, you know, we talked about how crypto is positively levered to the security developments in AI, but, you know, you can make the case that it's positively levered to most of the developments in the world right now.
例如，随着新商品和新智能的创造日益商品化，稀缺资产的价值反而会越来越高。
Like, for example, as as the creation of new as new creation of new goods and intelligence, etcetera, becomes commoditized, scarce assets become more valuable.
随着地缘政治不稳定加剧，那些更加主权化、脱离任何司法管辖区、类似于金融领域端到端加密的系统，将拥有更大的发展空间。
As geopolitical instability ensues, systems that are extra sovereign, right, outside of outside of any jurisdiction that that are kind of the equivalent of end to end encryption for finance, those have more space to thrive.
我觉得，我是在土耳其长大的。
And I think that, you know, I grew up in Turkey.
我大部分家人现在仍然在那里。
Most of my family is still there.
我觉得，那些在美国长大的人，或者一般在稳定世界中长大的人，都无法真正理解当世界变得不稳定时会发生什么。
I think that people who grew up in America do not have and in general, in sort of a stable world world, do not have the the sense for what can happen as the world destabilizes.
而且我觉得，我成长的那个国家中，越来越多的人开始接入加密货币体系，并将其作为救命稻草，这让我越来越清楚地看到，这项技术正处在一条复合增长的轨迹上，将产生巨大的影响。
And I think that, you know, as many people in in the country that I grew up in are starting to onboard the crypto rails and sort of using that as a lifeboat, I think it's increasingly clear to me that that this technology is is on a sort of compounding trajectory to do really massive things.
因此，结合这一点，再看看周围，竟然没人谈论它，这真的令人兴奋。
And so the combination of that, plus you look around and no one's even talking about it, it's just really exciting.
是的。
Yeah.
而且你似乎确信，人工智能的加速发展将利好加密货币，也就是说，会是水涨船高的局面。
And you do seem convinced that the acceleration of AI is going to benefit crypto, that it will be, you know, all boats rise together.
我认为，确实有一类软件行业，AI在短期内似乎并没有带来好处。
And I think that, well, there is some category of software industry that AI doesn't seem to benefit, at least in the short run.
你知道的。
You know?
Anthropic推出了一项新的安全模块，结果所有网络安全股票一天之内就下跌了10%到15%。
Anthropic drops a a new security module, and all the cybersecurity stocks drop, like, 10 to 15% in one day.
你为什么如此确信AI的加速发展会对加密货币有利？
Why are you so convinced that AI's acceleration will be beneficial to crypto?
这显然不是必然的，现在没有任何事情是确定的。
It's not obviously, nothing is guaranteed right now.
我认为，如果我们任由一切自然发展，加密货币可能会受到负面影响。
I think if we let everything run its course, it may be bad for crypto.
它也可能对加密货币有利。
It may be good for crypto.
我们还不知道。
We don't know.
我想我之所以有这种信念，是因为如果我们朝着我们希望的方向推动，就能让人工智能对加密货币产生极大的好处。
I guess the the the conviction that I have is that if we push things in the direction that we want them to go in, that we can make AI be extremely good for crypto.
而且，我认为这其中还有一个因素，就是如果你从第一性原理重新推导这一切，最终会到达一个与我们现在所处的加密货币状态非常相似的地方。
And and, also, I think there's the component of this where I do strongly believe that if you were to rederive all of this from first principles, you end up in a place that's very similar to where we currently landed with crypto.
而且，是的，我认为基于我们讨论过的所有原因，从根本上讲，加密货币对人工智能非常有利，人工智能对加密货币也非常有利。
And, yeah, I think I just think that for all the reasons we've talked about, that for fundamental reasons, crypto is extremely good for AI, and AI is extremely good for crypto.
所以我认为，你知道，没有什么是 guaranteed 的，我们仍然需要发挥自己的能动性。
So I I think that I think that, you know, nothing is guaranteed, and we still have to exercise our agency.
但基于我们迄今为止讨论的所有原因，对我来说，这些事物将以积极的方式融合在一起，这一点已经相当清楚了。
But but I think for all the reasons we've discussed so far, it's, like, pretty clear to me that that these things are gonna converge in a positive way.
那么，让我们以高度的能动性和坚定信念来结束今天的讨论吧，Alpin。
Well, let's end a note on high agency and conviction, Alpin.
非常感谢你今天加入我们。
Thank you so much for joining us today.
很好。
Cool.
谢谢你们邀请我。
Thanks for having me.
我得提醒一下Bankless的听众，当然，这一切都不是财务建议。
Gotta let you know, bankless listeners, of course, none of this has been financial advice.
你可能会损失你投入的资金。
You could lose what you put in.
希望那边有个大语言模型在保护它。
Hopefully, an LLM out there.
一位白帽黑客正在保护它。
A whitehat is protecting it.
我们正朝着西方前进。
We are headed west.
这就是前沿。
This is the frontier.
这不适合每个人，但我们很高兴你与我们一同踏上Bankless的旅程。
It's not for everyone, but we're glad you're with us on the bankless journey.
非常感谢。
Thanks a lot.
关于 Bayt 播客
Bayt 提供中文+原文双语音频和字幕，帮助你打破语言障碍，轻松听懂全球优质播客。