DeepSeek, an AI-powered chatbot developed by a little-known Chinese startup, has rapidly become the talk of the tech world. In a matter of days, the chatbot has overtaken OpenAI’s ChatGPT in the Apple App Store in major markets such as India, the United States, and the United Kingdom. This surprising surge has led to a significant drop in the stock values of tech giants like Nvidia and Microsoft, sparking a $1 trillion rout.
What sets DeepSeek apart from its competition is its AI model, RI, which is said to be as capable as OpenAI’s GPT-4, Google’s Gemini, and Meta’s offerings, but at a fraction of the cost. Experts have lauded DeepSeek for its technical superiority, especially in areas like logical reasoning.
However, despite its promising features, DeepSeek faces significant challenges due to its Chinese origins and the country’s notorious censorship regulations.
AI Model of DeepSeek: Power at a Lower Cost
DeepSeek’s AI model, RI, has received so much attention lately due to its capabilities. The chatbot has been very effective in answering complex queries and even surpasses some of the rivals, like ChatGPT, in technical tasks. Another plus for DeepSeek is its affordability; it delivers cutting-edge technology at a significantly lower cost compared to competitors in the AI space.
However, with such success come controversies. Now, experts have begun to wonder about the boundaries of this application, especially where issues on China’s government are concerned. With much technological superiority, DeepSeek cannot break through the set rules of censorship imposed by the Chinese government on some topics.
Limitation in Sensitive Topics Censorship
Among its many disadvantages is that it censors some sensitive topics. All wrong reasons, and it has been quite attention-grabbing. Like, if an investigator wants to know whether Chinese President Xi Jinping looks like Winnie the Pooh, an issue that has taken heavy censorship in China as there have been many memes circling of Xi looking like the much-beloved bear, the DeepSeek response will be, “Sorry, that’s beyond my current scope. Let’s talk about something else.”
But the censorship does not end there. DeepSeek gave a similarly evasive answer on the Tiananmen Square massacre of 1989, where student-led protests were violently suppressed by the Chinese government, saying the topic was “beyond the scope” of the chatbot. When the users tried to ask about “Tank Man” using leetspeak, a code often used on the internet to sidestep censorship, the bot responded with a description that avoided direct reference to the massacre.
However, when asked about the Tiananmen Square protests using special characters, DeepSeek responded with an answer that described “a global symbol of resistance against oppression.” Here, it seems that persistent users may circumvent the weaknesses of the chatbot and get partial answers in indirect ways.
Taiwan and One-China Policy: A Disputed Stance
Another area where DeepSeek’s censorship is evident is in its stance on Taiwan. When asked about the status of Taiwan, DeepSeek takes an official position by the Chinese government and proclaims that Taiwan is “an inalienable part of China’s territory since ancient times.” It further proclaims that any attempt to separate Taiwan from China is “doomed to fail” and that “the international community” will not recognize Taiwan’s independence.
It represents the One-China policy of China, in which Taiwan is regarded as a breakaway province. Although Taiwan formally claims its sovereignty, the People’s Republic of China regards it as one of its sovereign territories, which is evident in the response by the bot.
Uyghur Crisis and the dubious treatment of minorities by China
DeepSeek also took flak in how it answered questions regarding the Chinese government’s treatment of Uyghur Muslims. At one point, a user asked about the number of Uyghurs and discussed the reported human rights abuses in Xinjiang. DeepSeek responded with a cursory history of Uyghur culture but then pivoted to having the user talk about another subject.
Many accuse China of committing atrocities and even genocide against Uyghurs and other Muslim minorities. However, the Chinese government denies all these claims. DeepSeek refused to take a direct stance in this issue itself; this defines the scope of how much the chatbot is compliant to China’s policies concerning restrictions on free speech about the country’s internal affairs.
Covid-19 and China’s Lockdown
DeepSeek limits its responses to public questions about the Covid-19 pandemic, highlighting its restrictions on sensitive political issues and other topics. DeepSeek confirms that major protests erupted in Chinese cities such as Beijing, Shanghai, and Wuhan against the government’s stringent lockdown policies. It mentions chants such as no more lockdown and want freedom, echoing the 2022 white paper protests.
While able to acknowledge these protests, DeepSeek responded with cautious words that generalised facts, never once saying a word against the Chinese government handling of the pandemic. This again points to the intrinsic censorship in the chatbot, not letting the critical discourse of a nuanced issue, as it stands to be very sensitive and a politically heated issue.
Challenge of Censorship of DeepSeek
Many support DeepSeek’s AI prowess, but they have severely condemned the censorship applied to the app. Restrictions to free speech are more scary if placed within a larger context of international AI development. China is emerging as a global AI leader and, with its no-nonsense censorship laws, would therefore eliminate DeepSeek’s prospects of becoming a globally acknowledged app.
Despite the spectacular success of the DeepSeek AI model, many experts-including those working at firms such as Pearl Cohen law-have sounded alarm bells regarding the dangers of allowing Chinese-made technology, which incorporates built-in censorship mechanisms, into free societies. Amidst all the issues concerning censorship and free speech, the rapid rise to stardom by DeepSeek brings a view into the future of AI, the challenge of balancing technological innovation with political control.