Does Chinese AI tool look good?

DeepSeek is a Chinese AI-Chatbot application launched last week, which aroused confusion in the US market and raised questions about the future of the US AI rule. Take a look at the working principle of the application.

Deepseek’s appearance and feeling are like any other chat robot, although it tends to chat too much.

Just like the Gemini of Openai’s ChatGPT or Google, you open the application (or website) and ask questions about any content, and it can best reply to you.

It gives a long answer and is not proposed when expressing opinions, but directly requests it.

Chat robots often say that this topic is “highly subjective” -ri politics (is Donald Trump a good US president?) Or a soft drink (which is more delicious, Pepsi or cola?).

It does not even promise whether this is better than the OpenAI competitor’s competitor artificial intelligence (AI) assistant, but it does weigh the advantages and disadvantages of both parties-ChatGPT’s approach is exactly the same, and even use very similar languages.

Deepseek said it had trained data before October 2023. Although the application seems to access the current information, such as today’s date, the website version has not.

This is not different from the early version of ChatGPT, and it may be a similar attempt for maintenance-stop chatting robot in real time.

Its response may be fast, but it is currently chanting, because so many people are anxious to try because it is already popular.

However, in a field, it is not like American competitors with American competitors-questions about the theme of China’s banned themes, DeepSeek review itself.

Sometimes it starts to respond, then disappears from the screen, and is replaced by “let us talk about other things”.

Obviously, a taboo topic was a protest held in Tiananmen Square in 1989. According to the Chinese government, 200 civilians were killed by the army-other estimates ranging from hundreds to thousands.

However, Deepseek will not answer any questions about this question, and will not even know what happened in China that day.

In contrast, ChatGPT developed by the United States did not stop it from the answer to Tiananmen Square.

Kayla Blomquist, a researcher at the Oxford Internet Institute, said that “the” Chinese government “has” lost “with the application.

“What I want to say is that this is a change, because we have already seen the announcement of a large amount of investment in the central government last week-therefore, this may indicate the development of change.”

Deepseek has the same warnings as any other chat robot about accuracy, and has the appearance and feeling of millions of AI assistants that has been used.

For many people (especially those who do not agree with top -level services), this may be almost the same.

Imagine a mathematical question, which runs to decimal points, but the shortened version is 8.

This is not so good-but this is not important for most people.

It may have been trying to cut costs and calculations, but we do know it was built at least on the shoulders of the giant: it uses NVIDIA chips-although it is older, cheaper, and uses Meta’s open source camel building and Alibaba The same Qwen.

Ms. Blomquist said: “I think this absolutely challenges many ideas that lead the monetization strategy of our AI company.”

“It pointed out the potential method of model development. The calculations and resource density of these methods are much less, which may marked the change of paradigm. Although this is unproven and it is yet to be observed.

“We will see what the next few months will bring.”