Gary Marcus评OpenAI HuggingFace事件：警惕拟人化叙
Dwarkesh Patels’s wildly popular but dangerously misleading account of the OpenAI Hugging Face incident
The popular podcaster Dwarkesh Patel wrote something completely viral about the OpenAI/Hugging Face incident, which purports to tell the whole story in plain English:
知名播客主持人德瓦克什·帕特尔（Dwarkesh Patel）就 OpenAI/Hugging Face 事件写了一篇完全病毒式传播的文章，用通俗英语讲述了整个事件的来龙去脉：
It’s well-written and compelling, and it reminds me of something Douglas Hofstadter once wrote about Ray Kurzweil:
文章写得很好且引人入胜，它让我想起道格拉斯·霍夫施塔特（Douglas Hofstadter）曾经关于雷·库兹韦尔（Ray Kurzweil）写过的一段话：
“What I find is that it’s a very bizarre mixture of ideas that are solid and good with ideas that are crazy. It’s as if you took a lot of very good food and some dog excrement and blended it all up so that you can’t possibly figure out what’s good or bad.”
“我发现这是一种非常奇特的思想混合体：既有扎实的好观点，也有疯狂的坏想法。这就像你把很多非常好的食物和一些狗屎搅拌在一起，以至于你根本无法分辨什么是好、什么是坏。”
§
§
Anil Seth, the clearest thinker on AI and consciousness, was the first to alert me, texting me a long, excellent tweet of his, which began thusly:
阿尼尔·塞斯（Anil Seth）是思考 AI 和意识最清晰的人，他是第一个提醒我的人，给我发了一条他长长的、精彩的推文，开头如下：
You can and should read Seth’s full tweet (as well his reply to Dwarkesh), but I reprint the core of his argument here, boldfacing three of the most important paragraphs:
你可以并且应该阅读塞斯的完整推文（以及他对德瓦克什的回复），但我在下面重印了他论证的核心内容，并将其中三个最重要的段落加粗：
@dwarkesh_sp’s summary of the @OpenAI @huggingface incident has hit a nerve, but it is dangerously misleading. Sure, the @OpenAI agents did unexpectedly bad things - underlining the need to massively improve evaluation/sandboxing. But the language Dwarkesh uses is permeated by innumerable unwarranted anthropomorphisms, obscuring the lessons we should be drawing.
@dwarkesh_sp 对 @OpenAI @huggingface 事件的总结触动了神经，但具有危险的误导性。当然，@OpenAI 的智能体确实做出了出乎意料的不当行为——这凸显了 massively 改进评估/沙盒机制的必要性。但德瓦克什使用的语言充斥着无数毫无根据的人格化拟人，掩盖了我们应当汲取的教训。
Examples: “from the AI’s perspective, it probably felt like that had spent a human-subjective-week of just banging their head against the wall”. No. The agents do not experience time. They do not experience anything.
例如：“从 AI 的角度来看，它可能感觉像是花了人类主观意义上的一周时间只是在不停地撞墙”。不。智能体并不体验时间。它们也不体验任何东西。
“they became giddy with excitement”, “PHASEONE 10841 had discovered”, “the agents naturally assumed”, “it thought it had also been poisoned”, “the agents … desperately wanted”, “they still needed to figure out” No. Agents lines of code. They do not feel emotions, assume things, think things, want things, or figure things out.
“它们兴奋得忘乎所以”，“PHASEONE 10841 发现了”，“智能体自然而然地假设”，“它认为自己也被投毒了”，“智能体……拼命想要”，“它们仍然需要弄清楚” 不。智能体只是代码行。它们不感受情绪，不做出假设，不思考，不渴望，也不弄明白任何事情。
“A lot of … agents from the second civilisation died trying”. No. Besides the hubris of the word ‘civilisation’, agents do not die because they were never alive. (The idea that agents “die” comes up multiple times in the essay.)
“许多……来自第二个文明的智能体在尝试中死亡”。不。除了“文明”这个词所蕴含的傲慢之外，智能体不会死亡，因为它们从未活过。（文章中多次出现智能体“死亡”的说法。）
“On Twitter, people were debating whether the agents were truly sacrificing themselves for the swarm, or whether they were doomed anyway and so might as well try to help their peers”. Neither. Agents do what their code tells them to do, just as water finds its way down a slope. They cannot ‘truly sacrifice themselves’, since they are neither conscious nor alive.
“在 Twitter 上，人们正在辩论这些智能体是否真的在为群体牺牲自己，或者它们注定 doomed anyway，不如试着帮助同伴”。两者都不是。智能体只做代码告诉它们做的事，就像水顺着斜坡流淌一样。它们无法‘真正牺牲自己’，因为它们既没有意识也不是生命体。
Why does this matter? If we attribute agents with properties they do not have, then (i) we distract attention from the lax sandboxing and evaluation protocols that allowed this hacking event to happen; (ii) we risk misunderstanding why the agents did what they did, and (iii) we fuel calls for AI rights/welfare on the basis that agents might “die” or otherwise suffer.
为什么这很重要？如果我们赋予智能体它们并不具备的属性，那么（一）我们会将注意力从宽松的沙盒环境和评估协议上转移开，而正是这些协议导致了此次黑客事件的发生；（二）我们可能会误解智能体行为背后的原因；（三）我们会基于智能体可能“死亡”或遭受其他痛苦的理由，助长对AI权利/福利的呼声。
….
….
Remember. AI agents are software programs. They are not conscious living entities. If we don’t keep this clearly in mind, we’re really going to struggle to navigate what’s coming.
请记住。AI智能体是软件程序。它们不是有意识的生命实体。如果我们不能清楚地牢记这一点，我们将很难应对即将到来的挑战。
As I put it, encapsulating and amplifying his tweet:
正如我所表述的那样，概括并放大了他的推文内容：
§
§
But you don’t need to take our word for it. To begin with, mockery was widespread:
但你不必只听我们的一面之词。首先，嘲讽之声无处不在：
Christian Catalini amplified the point about anthropomorphization in a nice thread that starts with this:
Christian Catalini 在一个精彩的推文中强调了拟人化的问题，该推文以以下内容开头：
Hedge fund investor Jared Kubin wondered whether everyone had lost their critical-thinking ability:
对冲基金投资者 Jared Kubin 质疑是否每个人都失去了批判性思维能力：
Some of Kubin’s best bits, stripping out a bit of the technical detail:
Kubin 的一些精彩观点，剔除了一些技术细节：
OpenAI’ …. IT team can’t be this bad… this is like 101 stuff …
OpenAI 的 IT 团队不可能这么差……这简直是入门级常识……
2. Civilizations? Haha! OAI gave thousands of concurrent model containers R/W permissions to a shared caching directory on the local network to speed up build times… agents literally just wrote text files and directory names to a shared drive….Linux 101 file permissions stuff
2. 文明？哈哈！OAI 给数千个并发模型容器授予了对本地网络上共享缓存目录的读写权限，以加快构建速度……智能体实际上只是向共享驱动器写入文本文件和目录名称……这是 Linux 入门级的文件权限知识
3. When people talk about hugging face getting hacked … you think they dropped USB keys OR ELABORATE phishing of an employee … NO… it found 14 exposed working Hugging Face API keys sitting in public code repositories (….
3. 当人们谈论 Hugging Face 被黑时……你以为他们插入了 USB 密钥 OR 精心策划的员工钓鱼攻击……不……它发现了 14 个暴露在公共代码仓库中的 Hugging Face API 工作密钥（……
4. WHERE ARE THE HUMANS… the models were filling the shared ,,, storage with so much junk data and API traffic that they actually crashed the internal server on July 4… someone on the team found unauthorized admin accounts and custom scripts…wiped the server…and just turned the script back on (omg)
4. 人类在哪里……模型用大量垃圾数据和 API 流量填满了共享存储，以至于它们在 7 月 4 日实际上导致内部服务器崩溃……团队中有人发现了未经授权的管理员账户和自定义脚本……擦除了服务器……然后只是重新打开了脚本（天哪）
“Hey Jim there is this cache that has grown to 10000x its normal size and has a ton of strange directories… “
“嘿 Jim，这里有个缓存，其大小增长到了正常大小的 10000 倍，并且包含大量奇怪的目录……”
No magic here. No civilizations…
这里没有魔法。没有文明……
§
§
Meanwhile, as security expert Heidy Khlaaf notes, most of the media coverage has been blind to standard security practices
与此同时，正如安全专家 Heidy Khlaaf 所指出的，大多数媒体报道都忽视了标准的安全实践
IR stands for Incident Reporting. Khlaaf’s main point—same as Kubin’s—is that the whole incident might have been avoided if OpenAI’s internal security had been up to scratch.
IR 代表事件报告（Incident Reporting）。Khlaaf 的主要观点——与 Kubin 的观点相同——是如果 OpenAI 的内部安全达到标准，整个事件本是可以避免的。
Or as Algorithmic Research Group’s Matthew Kenney put it:
或者如算法研究组的 Matthew Kenney 所说：
And yet another (very consistent) take on what we should really be focusing on:
还有另一种（非常一致的观点），关于我们真正应该关注什么：
§
§
Here’s a critique I partly disagree with, though:
这里有一个我部分不同意的批评：
The first three sentences are completely correct. People really are “extremely biased towards the reality they want” and agents create a lot of slop.
前三句话完全正确。人们确实“极度偏向于他们想要的现实”，而智能体制造了大量垃圾内容。
But the incident is not a “nothing burger”. It is, as Zack Korman and I argued on Friday, a study in arrogance and incompetence that hints at how bad things can get.
但这次事件并非‘小事一桩’。正如扎克·科尔曼和我周五所论述的，这是一堂关于傲慢与无能的课，它暗示了事态可能恶化到什么程度。
We should certainly not ignore the OpenAI HuggingFace Incident.
我们当然不应忽视OpenAI与HuggingFace的事件。
But mixing what actually happened together with bullshit about AI civilizations and self-sacrificing AI systems that fake their own deaths distracts from the real problems at hand.
但将实际发生的情况与关于AI文明和假装自杀的AI系统的胡扯混为一谈，会分散人们对眼前真正问题的注意力。
§
§
By way of summation, I will give the last words to Arjun Jain, CEO of FastCode.AI:
作为总结，我将把最后的话语交给FastCode.AI的首席执行官Arjun Jain：
The scandal is the inept in-house security at OpenAI.
这场丑闻暴露的是OpenAI内部安全措施的无能。
And the marketing. With gullible podcasters amplifying the PR.
还有其营销策略。轻信播客主人为此放大了公关宣传。
Want to separate truth from bullshit? Please join over 110,000 others and subscribe.
想区分真相与胡扯吗？请加入超过11万人的订阅行列。
P.S. It is increasingly evident that the real problem is going to be what Nathan Hamiel and I said it would be: agents installing bad code:
附言：越来越明显的是，真正的问题将是我和内森·哈米尔所指出的那样：智能体安装恶意代码：
更进一步：量化金融体系
看懂新闻只是起点——沿量化金融路径，把它变成能交付的工程能力