Is the means of China so good it seems?

Deepseek, an AI-chatbot app, which began last week, has turned on chaos in the US markets and raised questions about the future of America’s dominance. Here, BBC technology editor Zoe Kleinman takes a look at the app.

Deepseek looks and feels like any other chatbot, though it bends towards being too clumsy.

As with Openai’s chatgt or Google Gemini, you open the app (or website) and ask them questions about anything, and it does its best to give you an answer.

It gives long answers and will not be drawn to the expression of an opinion, however one is directly required.

Chatbot often begins his response saying the topic is “very subjective” – ​​whether this is politics (is Donald Trump a good American president?) Or non -alcoholic beverages (which is more delicious, pepsi or Coke? ).

Nor would he commit to saying whether or not it was better than the artificial intelligence assistant (AI) of Openai, Chatgt, but it weighed the benefits and bad ones of both – chatgt did exactly the same, and even used very similar languages.

Deepseek says he was trained in data until October 2023, and while the app appears to have access to current information such as today, the website version is not.

This is not different with previous versions of the chatgpt and is probably a similar storage effort – to stop chatbot by distributing real -time internet pumped disinformation.

It may be fast enough in his answers, but is currently groaning under the weight of so many people who rush to try it after it has gone viral.

But there is an area in which it is nothing like its rivals in the US – Deepseek itself censors itself when it comes to questions about the forbidden subjects in China.

Sometimes an answer begins, which then disappears from the screen and is replaced by “Let’s talk about something else.”

An apparent taboo subject is the 1989 protests in Tiananmen Square which ended with 200 civilians killed by the military according to the Chinese government – other estimates have started from hundreds to thousands.

But Deepseek will not answer any questions about it, or even more widely about what happened in China that day.

Chatgpt developed by the US, for comparison, does not hold back to his responses about Tiananmen Square.

Kayla Blomquist, a researcher at the Oxford Internet Institute and director of Oxford China’s Oxford Policy Laboratory, says “relatively speaking” Chinese government has been “hands” with the app.

“I would say there is a difference after seeing a major investment announcement from the central government only in the last week – so it will probably signal a change moving forward.”

Deepseek comes with the same remarks as any other chatbots about accuracy, and has the appearance and feeling of the most determined US assistants already used by millions.

For many people – especially those who disagree on high -level services – maybe it feels very much.

Imagine a mathematical problem, in which the real response goes to 32 decimal places, but the abbreviated version goes to eight.

It’s not that good – but for most people, it will not matter.

It may be the case that he has managed to reduce costs and count, but we know that it is built at least partially on the shoulders of Giants: it uses nvidia chips – although older, cheaper versions – and uses the source open flawed Llama architecture as well as alibaba equivalent.

“I think this absolutely challenges the idea of ​​money gain strategies that have had many leadership firms,” ​​Ms Blomquist said.

“It is showing possible methods of developing model that are much less calculated and resources that will potentially signal a displacement in the paradigm, though this is unconfirmed and remains to be seen.

“We’ll see what they bring the next two months.”