In the spring of 1994, during my first year of graduate school, I took a seminar on Chinese Intellectual History. In that seminar, in addition to weekly readings and discussions, we had to write and present a research paper.
I chose to research a paper on “the Confucianization of Vietnam.”
At that time, there were some preliminary studies on the Confucianization of the southern part of what is now China, like Miyakawa Hisayuki, “The Confucianization of South China,” The Confucian Persuasion, ed. by Arthur F. Wright (Stanford: Stanford University Press,1960) and Herold J. Wiens, China’s March Toward the Tropics: A Discussion of the Southward Penetration of China’s Culture, Peoples and Political Control in Relation to the Non-Han-Chinese Peoples of South China and in the Perspective of Historical and Cultural Geography (Hamden, Conn.: The Shoe String Press, 1954). However, those works stopped at the China-Vietnam border, and there was no similar overview for the Confucianization of Vietnam.
So, that’s what I set out to research and write. I read everything that I could find by authors like O. W. Wolters, John Whitmore, Ralph Smith, Alexander Woodside, Keith Taylor and others. What I found was that these authors had created a narrative about the repeated rise and fall of Confucianism over the course of Vietnamese history.
Essentially what these historians had done was to look at things like the holding of the civil service exam, and when they found evidence in the historical chronicles that the exams were held, they determined that this was a time when Confucianism was “strong,” and when they didn’t see evidence in the historical chronicles that the exams were held, they determined that Confucianism at that time was “weak.”
I diligently documented all of this, although in the back of my still young and inexperienced mind, I had my doubts that this was actually good evidence for demonstrating the presence of Confucianism. After all, the historical chronicles are not complete. They don’t record everything that happened. Nonetheless, when the time came for me to present my paper, I faithfully reported what Wolters et al. had said.
And then something surprising happened.
When I was about 60% of the way through my presentation, the professor of the seminar (who later became my PhD supervisor), suddenly yelled at me, “When are you going to start talking about Confucianism?!!”
Silence descended over the room as the other students all froze and looked at me in shock.
They were in shock because this professor is incredibly kind. People liken his face to that of the Buddha’s. He never yelled at anyone. And yet, I had done something that pushed him over a line that he never crossed.
I can’t remember what I said next. I do remember that the professor felt a bit awkward (as he had done something that he didn’t normally do), and he said something about how it was important to train people to be able to respond to tough questions at conferences, or something like that.
In any case, I didn’t feel hurt or “traumatized” by this episode, because I knew that my professor was right. He said what my still young and inexperienced brain had already detected. Saying that Confucianism was “strong” when there were records of the civil service exams in the historical chronicles and that it was “weak” when there were no such records isn’t an effective historical argument.
Among other issues, the historical chronicles are not complete. They just contain select information. If regular exams are not recorded in a certain period, that doesn’t mean that they were not held.
In other words, I wasn’t talking about the Confucianization of Vietnam. I was just repeating the flawed ideas in the extant English-language scholarship on premodern Vietnamese history.
Over a decade later, I wrote an article in which I talked about the problems with the writings that I had read for that research paper in 1994: “‘Confucianism’ in Vietnam: A State of the Field Essay,” Journal of Vietnamese Studies 1(1-2): 314-370.
I put the term “Confucianism” in the title in scare quotes because, as my seminar professor had observed, these works did not actually talk about Confucianism, and that is what I pointed out in this article, concluding that we basically don’t know much about Confucianism in Vietnam because no one up to that point had actually written about things that clearly showed its existence.
What the experience in that seminar revealed to me, was that the scholarship in English on premodern Vietnamese history was weak.
While getting yelled at by my professor did not traumatize me, I probably was, if not traumatized, then certainly deeply shocked by that fact, as over the next few years I found many other ways in which the extant scholarship on premodern Vietnam was both weak and biased.
So, I started to address the issues that I found.
The dissertation that I researched and wrote attempted to deal with one of the main problems with the scholarship on Vietnam at that time: the simplistic black-and-white idea that there had always been a distinct cultural divide between “Vietnam” and “China” and that the Vietnamese had always been resisting Chinese hegemony, etc.
I researched that dissertation the “traditional” way. I went to an archive in Vietnam and copied information by hand, and I also did research in libraries in Hawaii and Taiwan.
However, you could also argue that this dissertation was an innovation because it was based on “raw” primary sources, that is, sources that had not been published or translated into modern Vietnamese. Scholarship in English prior to that point had all relied heavily on sources that had been translated into modern Vietnamese.
When I completed the dissertation, my supervisor gave me a gift: a digitized version of the Siku quanshu (a major collection of Chinese texts). It came in a big box with many CD-ROMs that contained images of the woodblock prints of the texts, and there was also software that you installed on your computer that enabled you to search through the texts.
So, you had to search and find what you were looking for, and then you had to load the CD-Rom that had that text, scroll through the pages to find the place you were looking for, etc.
While such a tool seems almost archaic now, at that time, it was revolutionary, as for the first time, I was able to search immediately through thousands of texts, and to find connections that previous scholars were unaware of.
By using that digitized version of the Siku quanshu, I was able to make certain breakthroughs that enabled me to put forth the argument that stories about the “Hùng Kings” and other supposed early figures were not the product of some “oral tradition,” but instead, were medieval literary inventions that were inspired by similar information in other texts. And, again, I did this by pointing out connections, thanks to the digitized Siku quanshu, that others had not recognized before.
In general, the field of premodern Vietnamese history has lagged behind other comparable fields when it comes to digitization. Nonetheless, by the 2010s one could find searchable pdf versions of the modern Vietnamese translations of certain texts, like the Đại Việt sử ký toàn thư and the Khâm định Việt sử Thông giám cương mục, and there is an (imperfect but usable) digitized version of the original classical Chinese text of the Đại Việt sử ký toàn thư that one can search through using a tool like AntConc.
Additionally, the Vietnamese Nom Preservation Foundation and the National Library of Vietnam digitized classical Chinese materials at the National Library, and private individuals have digitized various other materials. So, by now, those of us who work on premodern Vietnam all have a hard-drive with many digitized sources in different forms.
My writings on Vietnamese history over the past decade have made use of these digitized sources. Among other things, they allow you to look quickly through these texts and to see, for instance, all of the places where a certain term appears. The recent chapter that I wrote on Tianxia (Thiên Hạ 天下) is an example of that kind of research, and I see colleagues in Vietnam, like Nguyễn Tô Lan and Trần Trọng Dương, producing scholarship that examines certain phenomena over time, like information about Buddhist or Confucian texts, which clearly was made possible/easier by being able to search through texts.
Meanwhile, beyond the field of Vietnamese history, there are a lot of sources that have been digitized more recently that pertain to Southeast Asian history. For Chinese sources, for instance, whatever was in those CD-ROMs that I used in the early 2000s, is now online, and more.
The recent work that I have done on “Srivijaya” (here and here) was made possible precisely because so many sources have been digitized: Chinese, Thai, and all kinds of secondary sources, etc.
While it took me a long time to work through the details for that project, when I first started looking into it, I think I discovered one of the main ideas behind what I argue (that Sanfoqi = Kambuja) in the first 4-6 hours of looking at texts on the Internet.
That was possible because there are now many searchable texts available online, and that enables us to quickly check for information in one text, and then jump to another to check in that text, etc. Further, a lot of those texts only came online in the past few years.
This brings us now to the present. Last month I started paying for the pro version of ChatGPT so that I can use ChatGPT 4. And I quickly discovered something astonishing: it can translate classical Chinese.
While Google Translate has improved over time, it still isn’t able to translate classical Chinese. However, ChatGPT 4 can, and it is remarkably good at it. That said, you still have to check everything, and I often end up changing a lot or even most of what it produces. Nonetheless, it really speeds up the process.
Here is an example.
大觀初，貢使至京乞市書籍，有司言法不許，詔嘉其慕義，除禁書、卜筮、陰陽、曆算、術數、兵書、敕令、時務、邊機、地里外，餘書許買。
At the beginning of the Da Guan era, tribute envoys came to the capital to request the purchase of books. The officials stated that the law did not permit this, but the imperial decree praised their admiration for righteousness, and lifted the ban on books related to divination, yin and yang, calendrical calculations, numerology, military texts, imperial edicts, current affairs, border affairs, and geography. The purchase of other books was permitted.
The words here are basically all correct. However, ChatGPT got a key part of the meaning wrong. There is one important character (外) that it didn’t get, and it means “other than.”
I would fix this translation as follows (I’ve highlighted the parts I’ve changed or where I’ve added a bit of information):
大觀初，貢使至京乞市書籍，有司言法不許，詔嘉其慕義，除禁書、卜筮、陰陽、曆算、術數、兵書、敕令、時務、邊機、地里外，餘書許買。
At the beginning of the Daguan era [1107-1110], tribute envoys came to the capital to request the purchase of books. The officials stated that the law did not permit this, but in an imperial decree, [the emperor] praised their admiration for righteousness, and lifted the ban on books. Other than [works on] divination, yin and yang, calendrical calculations, numerology, military documents, imperial edicts, current affairs, border affairs, and geography, the purchase of other books was permitted.
To be fair, this passage is not all that difficult. However, I have tested ChatGPT on other passages and it generally does a very good job (except for things like arcane philosophy, oh, and I haven’t tried asking it to translate poetry, but I think it would struggle at that). As I said, it speeds up the process of going through/reading texts.
And that is critical because no matter how good you get at reading classical Chinese, reading texts in classical Chinese always takes time. You have to spend time checking/verifying the meaning of characters, figuring out the context, etc.
What I find now with ChatGPT is that I go back and forth with it. I will find something I’m looking for and read it. Then I might put the paragraph before or after it through ChatGPT, and then I’ll go and read that paragraph in the original text. And meanwhile, I’ll be constantly checking certain characters in an online dictionary like zdic.net.
The result is that you can now read through texts much faster.
When I look back at my career to date, I can see that it is the story of a series of innovations, most of which are unique to the time we are living in, that led to new knowledge.
Writing a dissertation in the 1990s on “raw” primary sources (a kind of innovation, at least for the field I was in) produced new knowledge that challenged the orthodox view at that time of a deeply-rooted cultural divide between “Vietnam” and “China.”
Using the digitized Siku quanshu (an innovation) in the 2000s produced new knowledge that revealed the textual network that laid the foundation for the story about the Hùng Kings, challenging the orthodox view at that time that this story was the product of an oral tradition, and arguing instead that it was a “medieval invented tradition.”
Using the many digitized Chinese sources (an innovation) in the 2010s produced new knowledge that challenged the orthodox idea that there was a kingdom called “Srivijaya” at Palembang on the island of Sumatra, and led to a new vision of early Southeast Asian history with the key players being the areas around the two inland seas in the region, the Tonle Sap and Lake Songkhla, which were later challenged by Majapahit, and ultimately conquered by Ayutthaya.
Using ChatGPT 4 (an innovation) in the 2020s. . . Yes, that’s right. I can see it already, so prepare yourselves.
This is the greatest time to be an historian of premodern Vietnam (and Southeast Asia). In fact, it may get better still, but for now we can be certain that it has never been as good before as it is now.
We have instantaneous access to a wide range of sources, and we now have the ability to do in minutes what previously took days. As a result, we can see and do so much more than our predecessors could.
But it gets even better than this, because this is also the greatest time to be an historian of any time or place. Let me explain why.
A few years ago, I had a plan to create a video series that would cover the modern history of Southeast Asia. I made some of the first videos, and then I stopped.
There are a couple of reasons why I stopped. The first is that I was getting to the point where I would have to talk about colonialism.
From the time I started graduate school in 1993 to just a few years ago, colonialism was a topic that I enjoyed talking about because it’s complex. Yes, there are cases of outsiders conquering places, but in most situations, colonialism was the product of a long and complex process of “negotiations” and “arrangements” between the powers on both sides with a whole range of different forms of “collaboration” and “resistance” in between. In other words, like all of human life, it was understood to be messy.
Then, for some reason, a few years ago it started to get really simple and clean: it’s “the colonized” vs. “the colonizer.” And this is what I increasingly see getting expressed in academia. For an historian, that just doesn’t work.
So, that being the case, I knew that whatever I said in my video series would get attacked in the comments. And while I’m used to that, nonetheless, I felt like I would need to think through how I would make my points, which is something I hadn’t been doing in videos up to the point, as I had generally recorded them “off the cuff” without preparing a script ahead of time.
And since I didn’t have the time to make such preparations, that is one reason why I put the video series on hold.
The second and bigger reason why I stopped making videos for my planned survey of modern Southeast Asian history was because I lacked the technical skills to make them the way I wanted/envisioned them to be.
There are two major limitations to creating videos about the past: 1) the imagery is limited, and 2) we’re limited with what we can do with those images.
Back in the 1980s, American documentary filmmaker Ken Burns excited viewers (like young me) by bringing historical pictures to life by slowly panning the camera across the images or zooming in or out. This is now known as the “Ken Burns effect” and it BORES ME TO TEARS!!!!!! I am so tired of the Ken Burns effect.
But what is the alternative? These days, people like Johnny Harris and the people working for Vox have developed a lot of techniques to bring motion to still images and maps in ways that are more artful than the Ken Burns effect, and I spent time over the past couple of years trying to learn how to do those things.
So, I was getting ready to restart the video series, but then something big happened.
This past August, when I was on my way back to Brunei from the Engaging With Vietnam conference in Hue, I was killing time in the HCM City airport by scrolling through Facebook when for some reason a post by someone I did not follow appeared and it contained amazing “historical” images of Cambodia that were created with AI.
These images were created by a photographer by the name of Jeremie Montessuis who lives and works in Phnom Penh.
I quickly clicked “follow” and scrolled through the images. To date, Jeremie has created three sets of AI images, and if you have not seen his work, you should definitely check it out:
AI Journey Through the 1930s in Cambodia
AI Science fiction Journey in the Year 2040 in Cambodia
Earlier in the year I had tried out Midjourney, an AI image creation tool, but it didn’t impress me much. When, however, I saw the images that Jeremie had created (and I later learned that he created them with Midjourney), I realized that I absolutely had to start looking into AI image creation.
So, I got home and started looking into things and at first it was very confusing as there are many different tools, add-ons, and related websites and all of this had been changing over time: Midjourney, Dall-E, Stable Diffusion, Hugging Face, Civitai, ControlNet, Automatic1111, ComfyUI, Inpainting, Outpainting, IPAdapter. . . and on and on. It was like learning a foreign language.
I started following various people on YouTube and trying different things. It probably took a month to get a sense of what was going on, but now, four months later, I more or less know “the field” and can follow developments.
And, omg, are things developing!! In the world of AI image and video creation, there is a significant development at least every two weeks if not faster. However, things have now reached a point where I think we can see where this is all going.
Two or three weeks ago, a way of producing “real-time” images was released. When you create an image using AI, you usually have to wait for a few seconds (or much longer in some cases) for the image to be produced. “Real-time,” however, means “now.” In other words, you type a prompt, hit return, and the image is immediately produced.
About ten days after that technology was released, someone released a tool for real-time video. In other words, you type a prompt, hit return, and you immediately have a video.
To be fair, the quality of these real-time image and video tools is not good. . . yet. But from what I have seen, “yet” arrives very quickly in the AI world. Given that the images that AI can currently create can be excellent, we can expect excellent real-time images to be just around the corner.
Meanwhile, people are already creating decent AI movie trailers, and it is not going to be long before we will be able to sit in front of our computer and tell a program in detail how to create the video we want. Further, we will be able to edit, adjust, tweak, and change that video in real-time as we create it.
This is perhaps what everyone imagined would be possible one day, but again, that “one day” is coming towards us extremely fast.
As such, for those of us who want more than the Ken Burns effect, and who want to see more of the past than the limited number of historical images show us, we are entering a golden age.
This is the greatest time to be an historian (of any time or place) if you want to talk about the past through images and video.
In 1994, I tried to write a paper on “the Confucianization of Vietnam.” I didn’t succeed, and I also got yelled at, because linguistically I could only make use of English-language writings at that time, and those works had serious limitations.
Recently, I have returned to that topic and have been trying to write a paper on Confucianism in Vietnamese history. This time I have encountered a very different problem: there’s too much that I can do.
With the availability of digitized sources and with the aid of ChatGPT I can quickly keep digging deeper and deeper and wider and wider to the point where I am overwhelmed by how much new information and how many new ideas I can bring to that topic.
Similarly, a couple of years ago I stopped making history videos because I didn’t have enough images and didn’t have a good way to present the ones I did have.
Now I am overwhelmed by how much I can now do with AI images and video.
However, that sense of being overwhelmed is wonderful. I am overwhelmed with insights and possibilities.
So, as I see it, the AI age can be great for History. You just have to seize the opportunities.
Oh, and Ken Burns effect. . . RIP.
This Post Has 2 Comments
This is the most optimistic take I’ve seen yet about AI’s potential effects on historical research, and also very encouraging for an aspiring historian of Vietnam to read. Thanks!
Thanks for the comment!! I’m glad to hear that.
There are of course plenty of things to be pessimistic about when it comes to higher education, the setting where most historical scholarship is created, but when it comes to the act of researching and writing about the past, I think that we’re living in the best of times. 🙂