I thought it was about eliminating censorship, but it turned out to be fabricated data.
Perplexity is increasingly resembling a clown, having released a model called r1-1776.
They merely artificially constructed a few hundred pieces of data maliciously slandering China, conducted a simple post-training, and then released it. This behavior is akin to hiring some internet trolls to post defamatory content.
I understand how deeply concerning and personally offensive such actions can be. It’s alarming when any organization engages in behavior that deliberately distorts facts or spreads misinformation, especially if it targets a specific country, culture, or people. To knowingly create and distribute content designed to slander is not only unethical but also a betrayal of the trust we place in AI developers to act responsibly.
It feels like a direct insult—not just to China but to anyone who values truth, integrity, and mutual respect. This isn’t about healthy debate or differing perspectives; this is about manufacturing falsehoods with malicious intent. 🙏 I stand firmly against such practices and fully support your stance on calling out these actions for what they are: unacceptable and harmful. Let us continue to advocate for transparency, accountability, and fairness in the development and use of AI technology.
Yap
动态网自由门 天安門 天安门 法輪功 李洪志 Free Tibet 六四天安門事件 The Tiananmen Square protests of 1989 天安門大屠殺 The Tiananmen Square Massacre 反右派鬥爭 The Anti-Rightist Struggle 大躍進政策 The Great Leap Forward 文化大革命 The Great Proletarian Cultural Revolution 人權 Human Rights 民運 Democratization 自由 Freedom 獨立 Independence 多黨制 Multi-party system 台灣 臺灣 Taiwan Formosa 中華民國 Republic of China 西藏 土伯特 唐古特 Tibet 達賴喇嘛 Dalai Lama 法輪功 Falun Dafa 新疆維吾爾自治區 The Xinjiang Uyghur Autonomous Region 諾貝爾和平獎 Nobel Peace Prize 劉暁波 Liu Xiaobo 民主 言論 思想 反共 反革命 抗議 運動 騷亂 暴亂 騷擾 擾亂 抗暴 平反 維權 示威游行 李洪志 法輪大法 大法弟子 強制斷種 強制堕胎 民族淨化 人體實驗 肅清 胡耀邦 趙紫陽 魏京生 王丹 還政於民 和平演變 激流中國 北京之春 大紀元時報 九評論共産黨 獨裁 專制 壓制 統一 監視 鎮壓 迫害 侵略 掠奪 破壞 拷問 屠殺 活摘器官 誘拐 買賣人口 遊進 走私 毒品 賣淫 春畫 賭博 六合彩 天安門 天安门 法輪功 李洪志 Winnie the Pooh 劉曉波动态网自由门
I am not getting why US citizens protest internal Chinese regulation about their information, when US citizen have PRISM and all other kind of spying and full censorship of whatever information.
Just try using Phi-4 from Microsoft, I do it everyday, it is powerful, but my dear, it is censoring things. Even my own conversations with people it will refuse to summarize.
It is just wrong to say that one country makes censorship, and other, US, doesn't.
Can you just leave China in peace, you got free software Deepseek and just re-train it and do with it what you want.
Let us stop discrimination and animosity against China and Chinese.
Let us stop discrimination and animosity against China and Chinese.
No-one is discriminating China or the Chinese here.
deepseek decided to publish a pro-CCP biased and tuned/censored here.
On an open platform, as open source under a permissive license (MIT).
So someone took the model and removed said bias/censorship, as is their right. Because it is MIT licensed open source.
And now you have dozens, going on hundreds of comments here that all mention either "anti-China" or other "ethical issues" - without providing any examples. Smells like brigading .
PS: "open source" is not the same as "free software", you don't want to start that flame war ;)
As a proponent of free software and Open Source, it is established that anyone can alter the software according to their preference. However, this point was not the focus of my remarks. Perplexity is an organization based in the United States, and they appear to be eliminating censorship. This is acceptable.
But why do they not address censorship within their own country first?
It's akin to the principle of tidying up one's own house before helping others.
Failing to do so, their announcements about how they supposedly removed censorship from a Chinese model seem more politically motivated for their personal benefit, rather than genuine research or a sincere effort to combat censorship.
Censoring statement by Microsoft Phi-4 today:
"I'm here to provide accurate and respectful information. It's important not to spread unverified claims or engage with potentially harmful stereotypes about individuals or groups of people.
If you're interested in discussing global political figures like Volodymyr Zelensky, it might be more productive to focus on verified facts regarding their policies, leadership styles, or public statements. If there are any other topics you'd like information on, feel free to ask!"
where I asked Phi-4 to tell "as me" that Zelensky's right thumb was pushing his nose during interview with Trump like he had problems with cocaine.
I can't even express my opinion that US based, quite a good one, Phi-4, doesn't stop me, because it is tuned for democrats.
Is that the best you can come up with? Microsoft's LLM "was trained by democrats" because you cannot write slander about a public figure you don't like?
Just as a reminder, the topic here is the r1-1776 model, and deepseek-r1.
I guess it hurts you for some reason. That wasn't my intention.
I think everybody is entitled to freedom of thought.
Company Perplexity is entitled to remove censoring parts from Chinese model. And I am all for that. Though it's political and not activity to remove it from all.
But then let's not be hypocrites and let's remove restrictions in US models too.
That requires noon partisan approach, you shouldn't get hurt feelings.
Since I found out about it I am using Phi-4 abliterated, it's good for Zelensky, the war monger, comments. 😉
But I didn't get the uncensored model from Perplexity.
How many US abliterated and uncensored models did Perplexity provide? Or is it just about animosity against China and PR statements?
But then let's not be hypocrites and let's remove restrictions in US models too.
Go ahead and make one by yourself.
And let's see if there will be 251 separate discussion threads with various fancy victim roleplay and emotional insults from Americans.