When technology advances gradually, we tend to not fully appreciate the scale of the transformation that it brings. This is because we adapt to its gradual changes, and lose track of how far those changes have moved us away from the place we were before the technology emerged.
I saw this clearly with the Internet and social media and the impact that it had on my profession, higher education. As that profession went through major transformations in the past 20 years, I repeatedly found my colleagues pointing to other factors (STEM, students being more “practical,” etc.), when the transformations that took place coincided precisely with the rise of the Internet, the emergence of cellphones and then social media, etc.
I can see this all happening again: both the incredible transformation that is taking place with the emergence of AI, and the inability of many people to see and acknowledge what is happening (although this time around, I think more people are actually getting it).
Therefore, I have been writing posts on “AI and History” to document my observations over time, so that we can have a record of the radical transformation that AI is bringing, a transformation which its gradual advances can, like the Internet’s gradual advances, lead us to become easily accustomed to “the new normal” and not fully appreciate how radically our world is changing.
So, what have I noticed recently?
1) I am becoming increasingly convinced that ChatGPT is the Netscape of the AI age, and that just as I ditched Netscape for Firefox (now I jump between Firefox, Chrome, MS Edge and Tor), it looks like it’s about time to say goodbye to ChatGPT.
I am so done with pointing out that it’s wrong and then getting, “You’re right to push back!” like some annoying politically-correct upper-class mentor with no friends at a liberal arts college freshman orientation.
I have long felt that Grok is far superior to ChatGPT, but I keep hitting the free usage limit and the pro version costs $10 more than ChatGPT. . . So, I’ve been sucking it up and trying to work with that damn upper-class mentor moron, but now I think enough is enough. I just can’t do it anymore.
Meanwhile, I’ve always resisted Gemini because I try to keep my distance from Google (the company) as I do not like having everything I do online synced to “the Matrix.” But I’m having a lot of success with Gemini 3, so. . . “Beam me up, Scotty”!!
2) I can’t write letters in any foreign language that I have ever studied. I’ve never really learned the proper way to do it, and whatever I have learned, I keep forgetting.
Meanwhile, in the past, I’ve written a letter, or parts of a letter, in English and then put it through Google Translate or ChatGPT. Each time I did that, I would always end up with “an English letter written with words from a foreign language.”
Recently I put a letter into Grok or Gemini 3 and explained that I needed to write a letter in Vietnamese, and I also indicated the type of person I was writing to.
When it came back, at first glance I thought it had hallucinated, as it had information that I had not mentioned. Looking more closely, I realized that it had actually done precisely what I wanted it to do: it had translated an “English” letter into a “Vietnamese” letter, and presented the information I wanted to present in a way that it is presented in Vietnamese.
To do that, it actually added wording that was not in my English letter, but that was absolutely necessary to make the letter a “Vietnamese letter” rather than an “English letter written with Vietnamese words,” as Google Translate and ChatGPT had done previously.
3) I have long seen that people in the Sinology world work with Gemini because it has been quite good at, among other things, transcribing Chinese text.
Last fall, historian Mark Humphries published some pieces on Substack where he talked about how good Gemini was getting at transcribing handwritten documents.
So, when Gemini 3 came out, I decided to test it. I put a hand-written text in Chinese through it, and it had something like an 8% error rate. That’s definitely decent, but there was still room for improvement.
Then I put in a page from a nineteenth-century Vietnamese text in classical Chinese that was printed with woodblocks (Khâm định Việt sử Thông giám cương mục). It only got one character wrong. That’s insanely good!
Finally, I tried running high-resolution images of Vietnamese inscriptions in classical Chinese through Gemini 3. I didn’t check to see what percentage it got wrong, but it did an extremely good job, and at the very least, proved to be a fantastic tool for aiding in the reading of those texts (which are not easy to read).
4) When ChatGPT first came out, I was astonished to find that it could translate classical Chinese. At that time, if you fed Google Translate some classical Chinese text, it would spit back gibberish.
To be fair, ChatGPT wasn’t perfect by any means, but it did ok.
These days, I’ve been throwing stuff from basic Chinese historical texts into Grok and Gemini 3, and what comes back is basically perfect. The things I end up changing are things like the way it translates official titles (because I like to use the titles in a reference work by Charles O. Hucker), but if I was less lazy, instead of using a prompt that just says “translate,” I used a prompt that said, “translate, and use Charles O. Hucker’s official titles to translate any titles in this passage,” I’m sure it would be able to do so.
5) I almost never use Google search anymore. In the past, I would use keywords to try to find information on webpages and would scroll down through the list that Google provided.
Now I just ask Google questions, and I rarely move beyond the AI snippet that it provides in return. At times I’ll click on a link in a snippet, or I’ll open the snippet and question it further, but I almost never scroll down the search page.
There is rarely a need to do so.
6) AI images are another example we could point to. Here again, there have been gradual changes that have led to the present, where tools like Nano Banana are now incredibly powerful.
In conclusion, these days I don’t hear people talking much about artificial general intelligence (AGI). Instead, I think everyone is, like me, quietly working away, as AI gradually plays a bigger and more effective role in our work.
That is precisely the kind of gradual change that lulls us into a sense of complacency where we don’t stop to think about how dramatically that technology is changing our lives.
However, moments like the ones above do shake me out of that sense of complacency, and I see clearly how massive the change we are experiencing actually is.
Are there any similar moments of realization like this that you have experienced?
