deepseek has censorship which makes it unreliable. But hopefully someone will fix that since its opensource. So at least there is that. google is garbage in comparison to chatgpt and i havent tried anything meta has produced since i dont want to have anything to do with that filth, so i dont know anything about that one.
So maybe there is some hope in long term, if things dont get worse elsewhere.
I asked the locally-hosted version about Tienanmen Square and it gave an accurate description, but the wording was a bit sympathetic to the CCP. It was like “according to some people, this is what happened, but the government contests this.”
deepseek has censorship which makes it unreliable. But hopefully someone will fix that since its opensource. So at least there is that. google is garbage in comparison to chatgpt and i havent tried anything meta has produced since i dont want to have anything to do with that filth, so i dont know anything about that one. So maybe there is some hope in long term, if things dont get worse elsewhere.
Deepseek only has censorship on their hosted site… if you run the open source model on your own hardware it doesn’t seem to censor
I asked the locally-hosted version about Tienanmen Square and it gave an accurate description, but the wording was a bit sympathetic to the CCP. It was like “according to some people, this is what happened, but the government contests this.”
Honestly that’s probably just the data it’s been trained on, rather than explicit instructions to censor the incident.
Makes sense that a Chinese AI would be trained on more Chinese sources than a western AI.
Chatgpt is also censored.
Oh, I forgot about Claude. Last time I tried it, it seemed on par or even better that ChatGPT-4o (but was missing features like browsing).