[rexpop]

> They want to argue with a machine about political issues or try to gossip with a llm

This perspective exhibits an extremely limited imagination. Perhaps I am using LLMs to populate my calendar from meeting minutes. Should the system choke on events adjacent to sensative political subjects? Will the LLM chuck the whole meeting if one person mentions Tiananmen, or perhaps something even more subtly transgressive of CCP's ideological position?

Any serious application risks running afoul of an invisible, unaccountable censor. Pre-emptive evasion of this censor will produce a chilling effect in which we anticipate the CCP's ideological priorities and habitually accommodate them. Essentially, we would be brainwashing ourselves.

Such was it like under Soviet occupation, as well. And such is it like under NSA surveillance. A chilling effect is devastating to the rights of the individual.

[tempeler]

You believe your llm are alive or not trained by a human. You do not look at it realistically. Do you think llm will you teach or find a way to crime? According to your idea, it should have no censorship; it has to do. I don't trust any human-made stuff. No one has the liability to tell the truth.

[rexpop]

> You believe your llm are alive or not trained by a human.

No, I do not.

I see your point, in fact. As the story goes, "In the days when Sussman was a novice..."

[tempeler]

This is an open-source tool. You can train and shape it however you want. You can teach it to behave like an SS soldier; you are free to do whatever you want. No one limits you. People forget that, or they bring their agendas. Therefore, no one cares what the other political views are; I can train it whatever I want.