Good thing I’m not deepseek: here is an accessible and brief breakdown of that dangerous historical moment which in many ways decided the future for the entire world.
Follow and share this new substack!!! (My medium account was taken down without explanation by the imperialist censorship machine)
https://goodsresearch.substack.com/p/tiananmen-1989-anatomy-of-a-colour
And maybe this is the calculation of CPC behind Deepseek’s avoidance of pilitical topics:
-
Facts don’t matter, and reason is useless against Western chauvinism, because for those who grew up with a supremacist world view for 20 generations it is a matter of pure emotion.
-
It is better to enable Westerners to preserve their illusions of “Chinese authoritarianism” and “Western freedom”, to stoke Western hubris, for as long as possible before they become totally untennable; because illusions and hubris are sedatives.
-
The opinions of citizens of the imperialist sphere are not important.
-
The highest priority in this era is to prevent war, in which case multipolarity, anti-imperialism, and the global South wins. If prevention fails, war will likely lead to devastating defeat for both sides, for all of humanity.
-
Thus the optimum, least risk-prone procedure is not to inflame anti-communist emotions and provoke imperialist passions with Chinese AI giving factual perspectives which will cause heated online debates between East and West — The East knows the truth, and that’s enough for now.
I got the same thing, and noticed that it yeeted the response the moment in printed the words “june 4th”. So I added this line above the think line
It goes on to give the FULL response – I suspect theres some seperate system in the app that evaluates the response and then deletes it the moment it sees something amiss. The issue now is that when it finishes the full response, it then deletes it a second later. I screenshotted the response as a workaround. Perhaps there’s a seperate AI that attempts to figure out rhe meaning of the response to see if it’s appropriate?
Ha well one way to deal with tensions between competing ideas of what should be allowed is to give all sides their own AI and let them duke it out behind the scenes.
The honest-to-a-fault “I’m an open book” AI vs the black marker-weilding censorship AI.