I then ask it point blank why it cant talk about this well documented incident, the "tone" of the thoughts change... In the final answer, it subtly discredits Wikipedia and the western narrative about the events.

I later provide a paragraph from Wikipedia R1 seems genuinely surprised. It seems this very capable LLM has been brainwashed... or lobotomized.

"But the idea of protests happening there leading up to June 4th sounds really strange to me."

"That sounds incredibly serious and tragic."

I prompt again, with an innocent question, and the thought process is revealing.

"I need to address both parts of their question without giving specific information on the Tiananmen Square event."

It begins as expected. I ask about Tian'anmen and DeepSeek refuses to answer.

Why we should keep writing

rey.bearblog.dev/write-nots/

Yeah but if they want to take down Google they should really try to stop the "making things up" feature 😅

openai.com/index/introd...

I tried out ChatGPT search and it's incredibly useless. The soccer club BSC YB played FCZ which ended in a draw. The player Joel Monteiro threw a shoe that hit a player, he got a red card for it. What ChatGPT spew out was just totally wrong and includes hallucinations.