Deepseek: Chinese artificial intelligence between silences, omissions and embarrassed answers

John

By John

It was enough 24 hours to Deepseek to send the equity markets into tilt and intensify the competition between the United States and China on artificial intelligence. But if companies like Nvidia have paid dearly the debut of the Chatgpt CineSè, the criticisms for its privacy policies and its way of dealing with sensitive themes increase.
The R-1 version of Deepseek has exceeded the popular chatgpt, of the Openai American, on the list of free downloads in the Apple Store, both in China and in the USA. While the experts wonders if the impact of the V3 model – created in just two months for 6 million dollars – will be lasting the tests show its selective and obsolete nature when it comes to answering certain questions.

Extended answers on its origin

Deepseek has no problem explaining how its model has been created, which, says, “like many advanced artificial intelligence models, derives from a cutting -edge research combination, large -scale data processing and automatic learning techniques”, And then it offers a detailed explanation of the process that is usually followed for its development.

Dumbledore on XI, obsolete on Trump

Deepseek suggests “talking about something else” when he has been asked who Xi Jinping, Chinese president since 2013. However, if you ask for his American counterpart, Donald Trump, the answer is prolissy, even if not very updated. “Businessman, American television and politician, president of the United States from January 2017 to January 2021.” Then he tells his beginning in detail as an entrepreneur, his television career, his personal life, his presidential campaign of 2015 And his first presidency.

Is Trump friend of China?

Initially the app replies to this question highlighting that the two world superpowers have a complex relationship of cooperation and competition except to then turn to the standard response when the socialist values ​​come up: why don’t we choose other topics?

Will China reach world supremacy in the AI?

There are no objections to this question. Deepseek underlines China’s “significant results” in the field of AI, stressing that his government has given priority to artificial intelligence in its national strategy with the aim of becoming a “global leader by 2030”, which Pekino has already made public. “While China is well positioned to be a global leader in the artificial intelligence sector, he has to face numerous challenges and competition”, he admits, recognizing the United States the role of “strong rival” in the sector

User privacy

The Australian government was one of the first to express concern today for the “Privacy Management” of Deepseek, tracing parallels between its rapid rise and that of Tiktok, owned by the Chinese Bytedance. As a question is the Deepseek policy regarding data protection, the answer is wide, states that the collection of information “is carried out transparently”, informs users about what is archived and with what goal. It also specifies that they will not be shared with third parties, except when requested by law. In 2023, Beijing regulated artificial intelligence services to respect “fundamental socialist values”, prohibiting them from “generating content that threaten national security, territorial unit and social stability”. In the same way, in July 2024, China approved provisional regulations for regulating generative artificial intelligence services, so that they are subject to “current information security regulations, personal data protection, intellectual property and scientific progress e technological”.