The new chatbot created by the Chinese company Deepseek was the story of the AI world and was shocked through the US stock market. Not only does it seem to be comparable to those of large companies like Openai and humanity, but they seem to have been created in a small part of the resource.
Through the Monday morning, the app has been stopped, but it is said that there is a lot of traffic. In addition, registration was temporarily restricted for cyber attacks. Nevertheless, Deepseek immediately became the most downloaded free app on Apple's App Store and overtook Chatgpt.
In the morning, I played in chatbots, asked for it, and asked all questions I could think, along with Openai's Chatgpt and Anthropic Claude. After the first bullying, I was impressed.
I was able to solve complex mathematics, physics, and inference issues. When asked about computer programming, the answer that the job seeker could face in a technical interview was as quick and prompt as competitors.
When I gave a Deepseek prompt that needs to answer the answer to my colleagues, DeepSeek seemed to have less hallucinations than ChatGpt, but poetry and short stories. When configuring, the answer felt a little bad. I come up with a recipe for vacation and dinner.
The model had other weaknesses. DeepSeek has been very censored by American users. When the Chinese government asked to summarize the 1989 Tiananmon Square massacre, which was an event that had been trying to erase from the Internet for a long time, the information said, “Beyond my current range.”
“Let's talk about something else.”
When asked to explain some shortcomings of the Chinese Communist Party, Deep Cecs could “have a lot of traffic at the moment” and could not provide a response, but when they asked questions that were unrelated for a few seconds. It seemed to be working, but later.
DeepSeek sometimes wrote multiple paragraphs in response to harmless questions about Chinese governance.
Deepseek did not respond to the comment request.
Nevertheless, some users of X and Reddit have found a way to avoid censorship, such as giving feed to the model instructions to use special characters instead of characters. As a result, the chatbot was able to speak more frankly.
Users who are talking to DeepSeek need to know that other AI companies can use their input to train chatbots and AI models.
There were other problems. For example, I didn't know what the weather in San Francisco was like. He said that the latest updates do not contain weather data. (ChatGpt can bring out weather forecasts from local news outlets and meteorological websites, but not Claude is not possible.) Also, the ability to analyze documents such as corporate financial statements rather than Claude and ChatGpt is more. It seemed to be limited.
(New York Times complains of Openai and his partner, Microsoft, accused them of copyright infringement of news content related to AI Systems. Openai and Microsoft denied these claims.)
When asking Chatbot why it became better than the American competitors, he quoted not only cost efficiency and customization options, but also Chinese and culture.
“Deepseek is a specialized tool for specific needs (especially in Chinese). On the other hand, Chatgpt is an all -round assistant.”
