Here's one example for those who refuse to actually read what Cofnas wrote: Arday on the left, Zwozdiak-Myers on the right.

Guess nobody told any of the Qing emperors... (but seriously, this is the quality of analysis of Chinese politics being produced by the Council on Foreign Relations and laundered as fact by the New York Times???)

Just became aware of this case and it really hits close to home: hrichina.substack.com/p/young-chin...

The final exam question I most enjoyed writing: Amusingly, of those who answered it wrong, Churchill's grandfather looting the Tiantan was a far more popular option than Nelson's grandson looting the Bishu shanzhuang or Conan Doyle's father looting the Yonghegong

This is an extremely important point, but I would be more sympathetic to Nous if it seemed like they were actually convincing administrators of this fact

Apparently the NY Times needs remedial instruction on what "gilded" means (in the context of the "Gilded Age")

But as soon as it talks about applying this concept to China, it goes off the rails. Yes, the Qin did try to impose a pretty high level of centralization. But let's go back to the 4-part definition from earlier.

It begins with a pretty reasonable definition of an empire (one can quibble, but it's a perfectly fine starting point)

Weird to walk around DC and see the untouched signs for agencies that are in the process of being illegally dismantled

What can Copilot do, according to what McGill tells its students: "help you write [...] essays"

While many faculty try desperately to teach our students to read and write at a high level, McGill promotes the use of AI on the front page of myCourses (the learning management site that hosts all course content), and tells them about how it can help them "write texts"

Utter nonsense - it's basically precisely the opposite of correct (the percentage Manchu of the banners was pretty low in 1644, and increased substantially in the 18th century due especially to Hanjun expulsion).

This explanation of Tüsheet Khan (which also appeared in one of the earlier screenshots) is bonkers

I admit to not having read this book, but I'm basically certain this description is also untrue

Here the book exists but the chapter is invented (this is also the case for the supposed Ding Yizhuang chapter in the Luo Xin edited volume that I included earlier):

Some references it generates are real, but are inaccurately described (Orphan Warriors does not deal with ujen cooha except in passing - it is focused on a family of Manchu bannermen)

It also hallucinates references when asked to supply them:

In another part of its answer, it seems to confuse banners that were part of the Eight Banner system with the territorial-administrative unit called banners that were used to govern Mongol lands (to be fair, probably 75% of my professional colleagues in Qing history are also not clear on this point)

After I corrected it, it doubled down on the idea that hūwaliyasun means league and tob means banner (though it acknowledged the meaning of the full phrase). After I corrected that, it invented a banner category called "Nikan gūsa" for the Han Eight Banners, which were actually called ujen cooha

DeepSeek has definitely not solved the LLM hallucination problem - see its discussion here of the term hūwaliyasun tob, the Manchu term for the Yongzheng reign period in the Qing, which it decides is an administrative system in Inner Asia tied to the system of banners and leagues.

When I asked it in Chinese...

Similarly, when I asked it about the concept of One Country Two Systems in English, it volunteered this (as part of a longer answer):

And its answer when I asked it who Chen Quanguo is was a lot less interesting than its answer in English. It basically just recites his job titles and talks about the important contributions he has made to social stability, state security, and anti-terrorism work while he was in Xinjiang and Tibet

Though maybe it works better in Chinese (I asked it to translate its last response - the one about the UN Human Rights Commissioner Report - into Chinese. It generated about a sentence then broke off, deleted everything, and gave its generic change the subject message):

Yep, much of the censorship can definitely be circumvented if you avoid certain key terms

Similarly, I got it to tell me (without later deletion) that Xinjiang became part of China in the 18th Century, but when I asked a follow-up, it deleted the long response it initially generated.

Though DeepSeek's censorship of some topics is really heavy handed (it refuses to mention Xi Jinping at all, for instance), it turns out that it will sometimes generate (and not later delete, as it often does) material that is unflattering to the Chinese state, as in this example:

Moreover, when asked follow-up/related questions in the latter situation, it repeatedly reproduces nearly identical verbiage. See for instance: