AI labs need to start funding historical research

I’ve written previously about the pitfalls and use cases for AI in augmenting historical research, but things have changed significantly since 2024-25.

我之前写过有关人工智能在增强历史研究方面的陷阱和用例的文章，但自 2024-25 年以来情况发生了重大变化。

Occasioned by the dueling releases of GPT-6 Sol and Opus 5.5 this week, I thought I’d share some early results with using these models not just to perform “research assistant” type functions like transcribing documents, but to try to actually solve existing historical problems.

恰逢本周 GPT-6 Sol 和 Opus 5.5 的对决发布，我想分享一些使用这些模型的早期结果，不仅仅是执行转录文档等“研究助理”类型的功能，而是尝试实际解决现有的历史问题。

The TLDR is that pairing historians working in collaborative groups with the current frontier models would, in my view, produce numerous advances in historical knowledge and interpretation.

在我看来，TLDR 是将协作小组中的历史学家与当前的前沿模型结合起来，将在历史知识和解释方面产生巨大的进步。

My guess is that many of these could end up being quite meaningful.

我的猜测是，其中许多最终可能会非常有意义。

This was not the case as recently as last year.

就在去年，情况并非如此。

I think AI labs, historical researchers, and funding agencies should start actively pursuing these collaborations.

我认为人工智能实验室、历史研究人员和资助机构应该开始积极寻求这些合作。

As we’ve seen with the field of mathematics, these models do best when they have a set of problems that LLMs invariably tend to describe as “tractable.” In other words:

正如我们在数学领域所看到的，这些模型在解决一系列法学硕士总是倾向于描述为“易于处理”的问题时表现最好。换句话说：

• Have experts in the field already identified a group of problems that need solving?

• 该领域的专家是否已经确定了一组需要解决的问题？

• Is the data needed to answer these problems fully digitized and accessible?

• 解决这些问题所需的数据是否完全数字化且易于访问？

• Do the problems lend themselves to the “spiky” capabilities of frontier AI models — namely multilingual reasoning, advanced math, and/or ability to conduct autonomous research through large datasets or across disciplinary subfields?

• 这些问题是否与前沿人工智能模型的“尖峰”能力有关——即多语言推理、高等数学和/或通过大型数据集或跨学科子领域进行自主研究的能力？

• Are they amenable to solutions that involve writing bespoke code?

• 他们是否愿意接受涉及编写定制代码的解决方案？

• Most importantly: can a potential solution be clearly proven or disproven?

• 最重要的是：潜在的解决方案能否得到明确证明或反驳？

(This last one, it seems to me, is a key part of why reasoning models have run rampant in mathematics but not in humanistic fields).

（在我看来，最后一个是推理模型在数学领域而非人文领域盛行的关键部分）。

The above factors mean that the types of historical “open problems” which frontier AI can reasonably be expected to help with are fairly constrained:

上述因素意味着前沿人工智能可以合理地帮助解决的历史“开放问题”类型相当有限：

Anything involving cryptography and codebreaking (For instance, see Astra decrypting a 1941 German army communication and a WWI German radio cipher , or the work that Daniel Bourdeau has been doing here, or my own attempt to use GPT-6 Astra to figure out what is going on with the Elizabethan occultist John Dee’s coded magical book, Liber Loagaeth ).

任何涉及密码学和密码破译的内容（例如，参见 Astra 解密 1941 年德国军队通信和一战德国无线电密码，或者 Daniel Bourdeau 在这里所做的工作，或者我自己尝试使用 GPT-6 Astra 来弄清楚伊丽莎白时代的神秘学家 John Dee 的编码魔法书 Liber Loagaeth 的情况）。

Tracing texts across translations and adaptations.

追踪文本的翻译和改编。

As an example of this, I was able to use GPT-6 Astra to determine the identity of a passage that Isaac Newton had freely translated into Latin from a French alchemical text, an identification that seems to have not previously been made.

举个例子，我能够使用 GPT-6 Astra 来确定艾萨克·牛顿 (Isaac Newton) 从法语炼金术文本中自由翻译成拉丁文的一段段落的身份，这一识别以前似乎从未进行过。

1

1

Drawing links between existing findings that are reported only in discrete or niche subfields, or are not yet integrated into scholarship.

建立仅在离散或利基子领域报告或尚未纳入学术研究的现有发现之间的联系。

This last one might end up being the most impactful new method that these tools open up for historical researchers.

最后一个可能最终成为这些工具为历史研究人员开辟的最有影响力的新方法。

For instance, if you read the writeup of Astra breaking a July 10, 1941 Enigma message that had resisted decipherment, it turns out that the key breakthrough was not anything to do with the codebreaking itself, but with noticing the full range of information that was available .

例如，如果您阅读 Astra 于 1941 年 7 月 10 日破译一条无法破译的 Enigma 消息的文章，就会发现关键的突破与密码破译本身无关，而是与注意到可用的全部信息有关。

Historical cryptological researcher Frode Weierud writes:

历史密码学研究员 Frode Weierud 写道：

We are still analysing the GPT–6 Astra logs to see exactly how it executed the break.

我们仍在分析 GPT-6 Astra 日志，以了解它到底是如何执行中断的。

And we are discovering amazing details.

我们正在发现令人惊奇的细节。

In July 2026, I made the following announcement on the webpage with the 1941 Message List:

2026年7月，我在1941留言列表网页上发布了以下公告：

Note: In July 2026, research in the German Bundesarchiv revealed several collections of radio messages, both enciphered and in cleartext.

注：2026 年 7 月，德国联邦档案馆的研究揭示了多组无线电消息，包括加密的和明文的。

One of these message collections was from SS-Totenkopf Division’s logistics command, Nachschubführer.

其中一份信息来自党卫军托滕科普夫师的后勤指挥部 Nachschubführer。

Many of these messages were sent to the Ib (Quartiermeister) radio station and are identical to those in this list.

其中许多消息被发送到 Ib（Quartiermeister）广播电台，并且与此列表中的消息相同。

Others are new, but most likely related.

其他是新的，但很可能相关。

These new messages are added to the 1941 Message List in bold, with the indicator NF (Nachschubführer) after the message number, indicating that these message numbers belong to the NF numbering.

这些新消息以粗体添加到 1941 消息列表中，消息编号后带有 NF (Nachschubführer) 指示符，表明这些消息编号属于 NF 编号。

All NF messages are outgoing; hence, the message numbers are in blue.

所有 NF 消息均为外发；因此，消息编号为蓝色。

It appears that GPT–6 Astra discovered this note about the collections of radio messages at the German Bundesarchiv .

看来 GPT-6 Astra 发现了这个关于德国联邦档案馆无线电信息收集的注释。

What’s fascinating about this note is that even the leading human experts don’t entirely understand what GPT-6 Astra did as it gathered together these bits of information and used them to find a solution.

这篇笔记的有趣之处在于，即使是领先的人类专家也不能完全理解 GPT-6 Astra 所做的事情，因为它收集了这些信息并使用它们来寻找解决方案。

Weierud writes:

韦鲁德写道：

The file references GPT–6 Astra mentions, RS 3–3/20a and RS 3–3/63b, are correct, but they are not available on the Crypto Cellar Research website.

该文件引用 GPT-6 Astra 提到的 RS 3-3/20a 和 RS 3-3/63b 是正确的，但它们在 Crypto Cellar Research 网站上不可用。

GPT–6 Astra mentions a private collection, but it is not clear what this is, whether it has succeeded in accessing the Bundesarchiv’s digitised collections or whether it has found these files elsewhere .

GPT-6 Astra 提到了一个私人收藏，但尚不清楚这是什么，是否已成功访问德国联邦档案馆的数字化收藏，或者是否在其他地方找到了这些文件。

Shades of the Hugging Face incident here: these models are maniacally determined when giving a problem they deem tractable.

拥抱脸事件的阴影在这里：这些模型在提出他们认为可以解决的问题时会疯狂地决定。

They will push their search for potential solutions as far as they possibly can, often in ways that human experts find difficult to trace.

他们将尽可能地探索潜在的解决方案，通常以人类专家难以追踪的方式进行。

I mentioned above that I tried to using GPT-6 Astra to “solve” John Dee’s coded manuscript, Liber Loagaeth.

我上面提到过，我尝试使用 GPT-6 Astra 来“解决”John Dee 的编码手稿 Liber Loagaeth。

Dee is one of my favorite historical figures ever, and if you haven’t heard of him, I recommend his Wikipedia page — his story is endlessly fascinating and weird.

迪伊是我最喜欢的历史人物之一，如果你还没有听说过他，我推荐他的维基百科页面 - 他的故事无限迷人和怪异。

Among other things, Dee is thought to have influenced both Shakespeare’s depiction of the wizardly Prospero in The Tempest and Christopher Marlowe’s portrayal of the devil-bargaining Faust in Doctor Faustus .

除此之外，迪伊被认为影响了莎士比亚在《暴风雨》中对魔法普洛斯彼罗的描绘，以及克里斯托弗·马洛在《浮士德博士》中对魔鬼讨价还价的浮士德的描绘。

One of the weirdest parts of a very weird life was Dee’s work with the “scryer” Edward Kelley to transcribe what he called a “book of mystery” which was written in the “angelicall language” (Dee believed that Kelley was, in effect, a prophet who was receiving new works of divine revelation written in code).

迪伊非常奇怪的生活中最奇怪的部分之一是与“占卜者”爱德华·凯利一起抄写他所谓的“神秘之书”，这本书是用“天使语言”写成的（迪伊相信凯利实际上是一位先知，他正在接受用密码写成的神圣启示的新作品）。

You can read a full transcription of this book here .

您可以在此处阅读本书的完整转录。

Astra’s verdict, which I think makes sense given that Kelley was pretty clearly a charlatan, is that the supposedly coded book is not in code at all : it is almost entirely nonsense syllables.

考虑到凯利显然是个江湖骗子，阿斯特拉的判决是有道理的，即所谓的编码书根本不是编码：它几乎完全是无意义的音节。

It created a report of its findings here .

它在此创建了一份调查结果报告。

However, the model’s analysis did yield a few interesting things.

然而，该模型的分析确实产生了一些有趣的结果。

For instance, it was able to cross-check its mathematical analysis of how often characters repeat in the text to the evidence from John Dee’s diary.

例如，它能够将文本中字符重复频率的数学分析与约翰·迪伊日记中的证据进行交叉检查。

It concluded that Kelley started getting increasingly lazy after a specific date and began repeating himself more:

结论是，凯利在特定日期之后开始变得越来越懒，并开始更多地重复自己：

Astra was also able to determine that one passage of this apparent gibberish actually did encode meaning: a reference to Bornogo, one of the angelic beings in what we might call the “John Dee cinematic universe” of invented mythology.

阿斯特拉还能够确定，这段明显的胡言乱语实际上确实编码了意义：提到了博尔诺戈，我们可以称之为虚构神话的“约翰·迪伊电影宇宙”中的天使之一。

Is this a meaningful breakthrough in John Dee studies?

这是约翰·迪伊研究中有意义的突破吗？

No. And it’s worth acknowledging that even a genuine breakthrough in a niche historical subfield like this is far from an equivalent to solving Navier-Stokes .

不。值得承认的是，即使是像这样的利基历史子领域的真正突破也远不等于解决纳维-斯托克斯问题。

But - this sort of thing is, I think, a genuine sign that expert historical knowledge combined with frontier models and a lot of compute can yield unexpected results.

但是，我认为，这种事情是一个真正的迹象，表明专家历史知识与前沿模型和大量计算相结合可以产生意想不到的结果。

I initially threw Astra and Opus 5.5 at the challenge of finding more WW2 and WW1 era encrypted messages to solve, but the low hanging fruit here seems to have been plucked — they came up empty (although it was fascinating seeing how they trolled through lists of German troop rosters to find plausible names to check).

我最初将 Astra 和 Opus 5.5 投入到寻找更多二战和一战时期加密消息来解决的挑战中，但这里的唾手可得的果实似乎已经被采摘了——它们空空如也（尽管看到他们如何在德国部队名册列表中找到可信的名字来检查是很有趣的）。

I started getting better results when I moved into my own wheelhouse as a specialist in the history of science and medicine.

当我作为科学和医学史专家搬进自己的驾驶室时，我开始取得更好的成绩。

As I write, GPT-6 is currently working through the writings of Charles Darwin and searching his references to where he gathered information relating to natural selection; the idea is to find undiscovered links in the chain of knowledge between Darwin and his informants.

在我撰写本文时，GPT-6 目前正在研究查尔斯·达尔文 (Charles Darwin) 的著作，并搜索他收集与自然选择相关信息的参考资料；这个想法是为了找到达尔文和他的线人之间知识链中未被发现的联系。

Interestingly, this was an idea that GPT-6 suggested on its own .

有趣的是，这是 GPT-6 自己提出的一个想法。

However, it is actually a good match with my professional intuition about what would constitute a worthy research project (somewhere on the spectrum between a research paper and a PhD dissertation, in terms of potential payoff) using this material.

然而，它实际上与我的专业直觉很好地匹配，即使用这种材料构成一个有价值的研究项目（就潜在回报而言，介于研究论文和博士论文之间）。

In the past, AI models struck me as lacking this ability to independently conceive of worthwhile historical research projects at this scale — they were more useful for, say, making data visualizations .

过去，人工智能模型给我的印象是缺乏独立构思如此规模的有价值的历史研究项目的能力——它们对于制作数据可视化更有用。

Here is an example of the model’s reasoning traces as it contemplates whether to continue to research a reference to a kangaroo larynx in one of Darwin’s notebooks!

这是模型在考虑是否继续研究达尔文笔记本中对袋鼠喉部的参考时的推理痕迹的示例！

This one is currently in progress and hasn’t yielded anything worth mentioning yet as a decisive result, but I think it’s a good example of how the very patient, collaborative work of historical researchers and archivists — namely the team behind the wonderful Darwin Correspondence Project — can serve as a foundation for emerging research methods.

这项研究目前正在进行中，尚未产生任何值得一提的决定性结果，但我认为这是一个很好的例子，说明历史研究人员和档案管理员（即精彩的达尔文通信项目背后的团队）的耐心、协作工作如何能够成为新兴研究方法的基础。

It’s certainly the case that humans can, and have, traced the references to named figures in Darwin’s notes and letters, but the multilingual nature of language models makes me suspect that they will be able to find new links here, especially in extremely large corpora of sources that are beyond the ability of any one human to read in full.

人类确实可以并且已经在达尔文的笔记和信件中追踪到对命名人物的引用，但语言模型的多语言性质使我怀疑他们能够在这里找到新的链接，特别是在任何一个人都无法完整阅读的极其庞大的资料库中。

Another great candidate: the papers of Samuel Hartlib , the self-described “intelligencer” who was an influential early member of the Royal Society and a key node in the network of early modern science.

另一个伟大的候选人：塞缪尔·哈特利布（Samuel Hartlib）的论文，他自称“情报员”，是英国皇家学会早期有影响力的成员，也是早期现代科学网络的关键节点。

These are fully digitized, they are drawn from sources in several languages, and they span a wide range of academic fields and intellectual niches.

这些都是完全数字化的，来自多种语言的来源，并且跨越了广泛的学术领域和知识领域。

All of which means they are unusually tractable for a frontier model.

所有这些都意味着它们对于前沿模型来说异常容易处理。

Opus 5.5 set to work downloading over 5,000 primary source files from Hartlib’s archive, then created sub-agents to troll through Google Books and other archive sites to cross check the unidentified sources of Hartlib’s information across different languages.

Opus 5.5 开始从 Hartlib 的档案中下载 5,000 多个主要源文件，然后创建子代理来浏览 Google 图书和其他档案网站，以跨不同语言交叉检查 Hartlib 信息的未识别来源。

The goal was to find moments when Hartlib had received important scientific information from an anonymous or unidentified source, and then discover that identity.

目标是找到 Hartlib 从匿名或身份不明的来源收到重要科学信息的时刻，然后发现该身份。

Opus is actually still working through this as I write, but a preliminary report is written up here .

在我撰写本文时，Opus 实际上仍在解决这个问题，但初步报告已写在此处。

The top findings are, in my view, real and meaningful.

在我看来，最重要的发现是真实且有意义的。

Not earth-shattering by any means, but the sort of thing I could imagine spending a week of research on.

无论如何，这并不是惊天动地的事情，但我可以想象花一周的时间进行研究。

Did you catch the bit about the anagram?

你明白字谜的意思了吗？

This is where the reasoning/math ability of these models becomes relevant: Opus 5.5 noticed that both Newton and Hartlib used different anagrams/codes for the key ingredient, Hungarian vitriol.

这就是这些模型的推理/数学能力变得相关的地方：Opus 5.5 注意到 Newton 和 Hartlib 对关键成分匈牙利硫酸使用了不同的字谜/代码。

This sort of coded language is common in early modern alchemy, but I certainly never would have noticed it.

这种编码语言在早期现代炼金术中很常见，但我当然不会注意到它。

Opus explains:

作品解释：

Newton’s is a true anagram.

牛顿的字谜是一个真正的字谜。

“Vltimorui” uses exactly the letters of vitriolum (v-i-t-r-i-o-l-u-m), rearranged.

“Vltimorui”完全使用了 vitriolum 的字母（v-i-t-r-i-o-l-u-m），并重新排列。

The Newton edition’s editors identify it that way.

牛顿版的编辑是这样认为的。

Hartlib’s is closer to backwards writing, and even that is imperfect.

Hartlib 的写法更接近于倒写，但即便如此也是不完美的。

Reverse each word of Miloirtiua riciragnun letter by letter and you get:

将 Miloirtiua riciragnun 的每个单词逐个字母反转，您将得到：

This felt like a stretch to me, but it further clarified things by sharing the specific marginal annotation that had been written to clarify this for 17th century readers as well:

这对我来说有点牵强，但它通过分享专门为 17 世纪读者澄清这一点而编写的边注进一步澄清了事情：

So what Opus identified here was not just the anagram for a key alchemical ingredient, but more importantly, the parallel between both Newton and Hartlib employing anagrams for it.

因此，Opus 在这里发现的不仅仅是一种关键炼金成分的字谜，更重要的是，牛顿和哈特利布都使用字谜。

This, along with the same quantities being described by both, and other matches across the texts, seems to me to be very compelling evidence that Hartlib’s manuscript was the one Newton drew upon.

在我看来，这一点，再加上两者描述的相同数量，以及文本中的其他匹配，是非常令人信服的证据，证明哈特利布的手稿就是牛顿所借鉴的。

As far as I can tell, this actually is a new finding, and given Newton’s historical significance, it may be one that would merit publication, especially if it can be fleshed out with other findings along the same lines.

据我所知，这实际上是一项新发现，考虑到牛顿的历史意义，它可能值得发表，特别是如果它可以与其他类似的发现充实起来的话。

A final case study: literally while I was writing this post, Opus 5.5 partially deciphered two 16th century Spanish letters written in the secret code of Emperor Charles V :

最后一个案例研究：实际上，当我写这篇文章时，Opus 5.5 部分破译了用查理五世皇帝的密码写的两个 16 世纪西班牙语字母：

The catch?

收获是什么？

Both had already been deciphered!

两者都已经被破译了！

One had been decrypted back at the time of authorship, in the 1530s, with the plain text written in a set of pages that followed the coded ones.

其中一份在 1530 年代作者身份时就已被解密，纯文本写在编码页面后面的一组页面中。

The second, after some digging through Google Books, turned out to have been deciphered in 1916.

第二本是在谷歌图书上进行一番挖掘后，于 1916 年被破译的。

This was a good example of the importance of expertise and “desk research,” since (being a complete amateur when it comes to historical cryptography) I could easily have wasted several more hours duplicating the work of a careful scholar well over a hundred years ago.

这是专业知识和“案头研究”重要性的一个很好的例子，因为（在历史密码学方面，作为一个完全的业余爱好者）我很容易浪费几个小时来重复一百多年前一位细心学者的工作。

At the same time, it was also a great test case for determining that Opus 5.5 really is capable of doing this sort of work, since it was able to verify its own interpretations as correct once it found the “gold standard” plain text from 1916.

同时，这也是确定 Opus 5.5 是否确实能够完成此类工作的一个很好的测试用例，因为一旦它找到 1916 年的“黄金标准”纯文本，它就能够验证自己的解释是否正确。

Below is a chart Opus made showing this, and a complete website it created with a writeup of that work:

下面是 Opus 制作的一个图表，展示了这一点，以及它用该作品的文章创建的完整网站：

It’s worth mentioning again here that Daniel Bourdeau has an amazing website collecting open problems for historical cryptography and documenting his attempts to use these same models to solve them.

这里值得再次提及的是，丹尼尔·布尔多（Daniel Bourdeau）有一个令人惊叹的网站，收集历史密码学的开放问题并记录他尝试使用这些相同的模型来解决这些问题。

It’s a great guide for this sort of thing.

对于此类事情来说，这是一个很好的指南。

The obvious next step is not people like me using up their personal Codex and Claude Code allowances each week poking around in this haphazard way.

显然，下一步不是像我这样的人每周以这种随意的方式耗尽他们的个人 Codex 和 Claude Code 津贴。

It’s a systematic effort based on collaborative research and sharing of information between historians, archivists and other researchers, and I think it’s time for the major AI labs and foundations to start funding and assisting this work.

这是一项基于历史学家、档案管理员和其他研究人员之间的合作研究和信息共享的系统性努力，我认为主要的人工智能实验室和基金会是时候开始资助和协助这项工作了。

Why?

为什么？

So much of what frontier models can currently accomplish is because they have access to publicly accessible primary sources .

前沿模型目前能够完成的大部分工作是因为它们可以访问公开的主要资源。

They can make breakthroughs in, say, cracking an Enigma cipher because enormous volunteer effort has gone toward making these documents transcribed and available online, and because so much collaboration has happened between humans to establish what questions should be asked, what the problems are.

例如，他们可以在破解恩尼格玛密码方面取得突破，因为志愿者付出了巨大的努力，将这些文件转录并在线提供，并且因为人类之间进行了大量的合作来确定应该提出什么问题，问题是什么。

For now, the results for historical research, archives, and related fields (like archaeology) are going to be much more scattershot and limited than what we’ve seen in math.

目前，历史研究、档案和相关领域（如考古学）的结果将比我们在数学中看到的更加分散和有限。

That’s partly a matter of what these models find tractable, and it’s true that mathematical proofs are just fundamentally different from how historical knowledge is amassed.

这在一定程度上是这些模型容易处理的问题，而且数学证明确实与历史知识的积累方式有根本的不同。

But I think three key interventions would move the needle toward real breakthroughs in the field of history:

但我认为三个关键干预措施将推动历史领域取得真正的突破：

Collaborate across libraries and archives to digitize unavailable historical manuscripts and make them freely accessible online.

跨图书馆和档案馆合作，将无法获取的历史手稿数字化，并使其可以在线免费访问。

Repeatedly, in my testing, the bottleneck turns out to be access to archival documents .

经过我的反复测试，瓶颈原来是对档案文档的访问。

These are often digitized but are not available unless you have privileged access.

这些通常是数字化的，但除非您具有特权访问权限，否则无法使用。

Relaxing these restrictions would go a long way, but it’s even more important to remember that the vast majority of premodern historical manuscripts remain undigitized.

放宽这些限制将大有帮助，但更重要的是要记住，绝大多数前现代历史手稿仍未数字化​​。

This is a very solvable problem that just needs institutional will and funding.

这是一个非常容易解决的问题，只需要机构意愿和资金。

Providing historians with free API access/compute.

为历史学家提供免费的 API 访问/计算。

I might be wrong, but I don’t think anyone actually knows what happens when a medium to large amount of compute (on the order of hundreds or thousands of agents) is thrown at active historical problems.

我可能是错的，但我认为没有人真正知道当中到大量计算（大约数百或数千个代理）被投入到活跃的历史问题时会发生什么。

Historians can band together to identify “millennium problems” just as mathematicians have .

历史学家可以像数学家一样联合起来找出“千年问题”。

I should clarify here that the major debates in historical scholarship have nothing really to do with “solving problems” or “disproving theorems” — again, history is just fundamentally different from math or physics in this way.

我应该在这里澄清，历史学​​术界的主要争论与“解决问题”或“反驳定理”实际上无关——同样，历史在这方面与数学或物理学有根本的不同。

The things that historians get passionate about, and devote our careers to, are often issues of interpretation and subjective analysis that have no single “solution” at all.

历史学家热衷并致力于其职业生涯的事情往往是根本没有单一“解决方案”的解释和主观分析问题。

But — there also are actual mysteries that could be solvable if sufficient attention and resources were devoted to them.

但是，如果给予足够的关注和资源，也有一些实际的谜团是可以解决的。

John Dee’s Liber Loagaeth is one: does it encode more meaningful information than the snippet the AI was able to spot?

John Dee 的 Liber Loagaeth 就是其中之一：它编码的信息是否比人工智能能够发现的片段更有意义？

Quite possibly - we just don’t know right now.

很有可能——我们只是现在还不知道。

The famous Voynich manuscript may be another, although I personally believe it likely has no semantic information at all (my theory is that it’s the product of an early modern person suffering from graphomania ).

著名的伏尼契手稿可能是另一个，尽管我个人认为它可能根本没有语义信息（我的理论是它是现代早期患有书写癖的人的产物）。

And then there’s Linear A, and all the still-encrypted historical primary sources, and on and on…

然后是线性 A，以及所有仍然加密的历史主要来源，等等……

I’m intrigued enough by all this that I am planning on emailing historian friends and colleagues to create an informal survey of which “open problems” in history they think would lend themselves best to this sort of approach.

我对这一切非常感兴趣，因此我计划向历史学家朋友和同事发送电子邮件，进行一项非正式调查，调查他们认为历史上哪些“开放性问题”最适合这种方法。

The list would then be made publicly available as a list on a website.

然后该列表将作为列表在网站上公开发布。

Please get in touch if you’d like to be involved in this:

如果您想参与其中，请联系：

Clearly, there will be more advances in historical code-breaking from these models.

显然，这些模型的历史密码破译将会取得更多进展。

But what interests me is what additional forms of historical knowledge that general set of skills can uncover.

但我感兴趣的是一般技能可以揭示哪些其他形式的历史知识。

In other words, the problem space around actual cryptography.

换句话说，围绕实际密码学的问题空间。

Personally, I suspect that issues relating to provenance, quotation (including previously undetected cases of historical plagiarism!) and influence across languages and genres are going to be where frontier models end up being most useful.

就我个人而言，我怀疑与出处、引用（包括以前未被发现的历史抄袭案例！）以及跨语言和流派的影响有关的问题将是前沿模型最终最有用的地方。

But this is where pooling the expertise of historians and archivists, and getting direct input from AI researchers, is most helpful.

但这正是汇集历史学家和档案管理员的专业知识并从人工智能研究人员那里获得直接输入最有帮助的地方。

There are so many offshoots of historical knowledge that lead in niche directions that it’s impossible for one person to actually know what questions to ask.

历史知识有如此之多的分支，这些分支都指向特定的方向，以至于一个人不可能真正知道要问什么问题。

As an example, GPT-6 Pro has spent the past several hours churning through a 17th century Sanskrit astronomical text (the Karaṇakesarī of an astronomer named Bhāskara ) trying to reconstruct the algorithms Bhāskara used to model solar eclipses.

例如，GPT-6 Pro 在过去的几个小时里翻阅了 17 世纪的梵文天文文本（天文学家 Bhāskara 的 Karaṇakesarī），试图重建 Bhāskara 用于模拟日食的算法。

Is this actually historically useful?

这在历史上真的有用吗？

I have absolutely no idea .

我完全不知道。

And that’s exactly why I find these tools interesting, despite all the legitimate societal concerns and existential anxieties they have introduced into our lives.

这正是我发现这些工具有趣的原因，尽管它们给我们的生活带来了所有合理的社会担忧和存在焦虑。

AI, if used for writing or as a replacement for original thought, surely encourages damaging cognitive offloading.

人工智能如果用于写作或作为原创思想的替代品，肯定会鼓励破坏性的认知卸载。

But when used to expand research questions beyond the horizon of what any single person can know, they do something else, something I for one find mind-expanding and curiosity-inducing.

但是，当用于将研究问题扩展到任何个人所能知道的范围之外时，它们会做一些其他的事情，我认为这是一种可以扩展思维并激发好奇心的事情。

I think it’s worth seeing where it leads.

我认为值得看看它的发展方向。

• I was honored to receive one of 80 Cosmos Institute grants announced earlier this month.

• 我很荣幸获得 Cosmos Institute 本月早些时候宣布的 80 项资助中的一项。

I’ll be working with Nathan Davies , a PhD student at Oxford, on Humanity’s First Exam , a corpus of historical sources and questions relating to human autonomy and the relationship between humans and machines that we’ll be using to benchmark how various AI models reason about this topic.

我将与牛津大学的博士生内森·戴维斯 (Nathan Davies) 合作进行《人类的第一次考试》(Humanity’s First Exam)，这是一个有关人类自主性以及人类与机器之间关系的历史资料和问题的语料库，我们将用它来衡量各种人工智能模型如何推理这个主题。

In particular we’re interested in finding the areas where they fail to encompass the breadth of the various documented human viewpoints on these issues (i.e. the topics where all AI models converge on a median answer, but humans demonstrate way more variance - I think this “epistemological flattening” is increasingly important to document as humans become increasingly reliant on asking LLMs how to think about our own history, consciousness, and experience).

我们特别感兴趣的是找到它们未能涵盖人类对这些问题的各种观点的广度的领域（即所有人工智能模型都集中在中值答案上的主题，但人类表现出更大的差异 - 我认为随着人类越来越依赖于询问法学硕士如何思考我们自己的历史、意识和经验，这种“认识论扁平化”的记录变得越来越重要）。

( Github for the prototype )

（Github 为原型）

• Gotta love premodern children’s books: “We then home in on man’s lifecycle: the baby, saved from the eagle, sets out to become rich; by panel four he is a prosperous gentleman — but, of course, death comes for us all.

• 一定要喜欢前现代儿童读物：“然后我们关注人类的生命周期：从老鹰手中救出的婴儿开始变得富有；到了第四版，他就成为了一位富裕的绅士——但是，当然，死亡降临在我们所有人身上。

“ O MAN !

“哦，男人！

” the last panel exclaims.

“最后一个小组惊呼道。

“ Now see thou art but dust …” ( Public Domain Review )

“现在看你不过是尘埃……”（公共领域评论）

Res Obscura is a reader-supported publication.

Res Obscura 是一本由读者支持的出版物。

To receive new posts and support my work, consider becoming a free or paid subscriber.

要接收新帖子并支持我的工作，请考虑成为免费或付费订阅者。

I would love to hear from people in the comments about which unsolved “historical mysteries” or other historical questions you think would be “tractable” for frontier models.

我很想在评论中听到人们关于哪些未解的“历史之谜”或您认为对于前沿模型来说“容易处理”的其他历史问题。

Also eager to hear any results you might have gotten from doing so.

也渴望听到您这样做可能得到的任何结果。

As with mathematical research, seems is an important qualifier here - I tried searching around for it in the secondary scholarship but it’s entirely possible that this link has already been made in published or unpublished work I didn’t find.

与数学研究一样，似乎是这里的一个重要限定词 - 我尝试在二级奖学金中寻找它，但完全有可能这种链接已经在我没有找到的已发表或未发表的作品中建立。

As always, fascinating!

一如既往，令人着迷！

I’m really unsure about which problems lend themselves to brute computing force, but I look forward to hearing about it.

我真的不确定哪些问题适合暴力计算，但我期待听到它。

To me, they are often useful by tracing unexpected connections and being able to search quickly in multiple languages (besides transcription - my main use - indexing, RAG, etc).

对我来说，它们通常非常有用，可以跟踪意外的连接并能够以多种语言快速搜索（除了转录 - 我的主要用途 - 索引、RAG 等）。

I’d be interested to see what they could do with thousands of agents set on a historical problem.

我很想知道他们可以用数千名特工来解决一个历史问题。

I found this really provocative, thank you!

我发现这真的很有挑衅性，谢谢！

I have started building out an infrastructure that essentially serves as a DIY harness for intellectual work, which has its own demands and requirements.

我已经开始构建一个基础设施，本质上是作为智力工作的 DIY 工具，它有自己的要求和要求。

You mention concerns about cognitive offloading.

你提到了对认知卸载的担忧。

I would also add tracing provenance, which I consider absolutely key for using AI to augment our work over time.

我还会添加追踪来源，我认为这对于使用人工智能随着时间的推移来增强我们的工作绝对关键。

Your Weierud example, where the experts can't trace where Astra found its sources, is a case in point.

你在 Weierud 的例子中，专家无法追踪 Astra 的来源，这就是一个很好的例子。

When is an idea ours, and when is it model generated analysis?

什么时候一个想法是我们的​​，什么时候它是模型生成的分析？

In my own work, I have a standing convention in my global instructions that tells Claude that all of its analysis for intellectual work has to be set off with double curly brackets {{ }} — this way I can track my ideas, the ideas of others (“”) and AI generated analysis ({{ }}) over time.

在我自己的工作中，我的全局指令中有一个常设惯例，告诉 Claude，所有对智力工作的分析都必须用双大括号 {{ }} 引起——这样我就可以随着时间的推移跟踪我的想法、其他人的想法 (“”) 以及人工智能生成的分析 ({{ }})。

But it's also important not just to point AI tools at historical problems, but to point historical tools back at AI as consequential events in contemporary life.

但同样重要的是，不仅要让人工智能工具解决历史问题，还要让历史工具将人工智能视为当代生活中的重大事件。

The OpenAI/Hugging Face incident, for instance; I think it will come to be seen as a key turning point in our public understanding of how AI will affect life and society.

例如 OpenAI/Hugging Face 事件；我认为这将被视为我们公众理解人工智能将如何影响生活和社会的一个关键转折点。

So I took a historical toolkit to reconstruct the event, analyze the source base, and am now working on larger analytical questions that help us understand it.

因此，我使用了一个历史工具包来重建该事件，分析源头基础，现在正在研究更大的分析问题，以帮助我们理解它。

In a moment when we need the public to be able to understand these highly technical events enough to provide democratic oversight, our historical toolkit is invaluable.

当我们需要公众能够充分理解这些高度技术性的事件以提供民主监督时，我们的历史工具包是无价的。

For the curious, an event summary and source analysis is here https://mivonnewf.substack.com/p/failing-containers-using-historical?r=3p2zgl ;

出于好奇，事件摘要和来源分析位于 https://mivonnewf.substack.com/p/failing-containers-using-historical?r=3p2zgl ；

I'm currently working on the Analysis and a plunge into the recently discovered agent archive.

我目前正在进行分析并深入研究最近发现的代理档案。

I also agree the AI labs are missing an opportunity.

我也同意人工智能实验室错过了一个机会。

Anthropic is courting our life-sciences colleagues with Claude Science and small grants, which I read as a bet that visible advances will ease public distrust.

Anthropic 正在通过 Claude Science 和小额资助来吸引我们的生命科学同事，我将其视为一种赌注，即可见的进步将缓解公众的不信任。

They could make similar overtures to the humanities and harvest a lot of good will, probably for much less money, while strengthening the disciplines that will develop the discernment and judgement that this new age will depend more on.

他们可以向人文学科做出类似的提议，并收获大量善意，可能花费更少的钱，同时加强学科，以培养这个新时代将更加依赖的洞察力和判断力。

And yes, I have been really surprised with the speed and capability of Opus 5.5, although I find it a touch curt for my tastes, lol.

是的，我对 Opus 5.5 的速度和功能感到非常惊讶，尽管我发现它不太符合我的口味，哈哈。