1 week ago
10
You have a preview view of this article while we are checking your access. When we have confirmed access, the full article content will load.
The chatbot from China appears to perform a number of tasks as well as its American competitors do, but it censors topics such as Tiananmen Square.
By Eli Tan
Reporting from San Francisco
Jan. 27, 2025, 6:33 p.m. ET
A new chatbot created by the Chinese company DeepSeek is the talk of the A.I. world, and has sent a jolt through the U.S. stock market. Not only does it seem to be comparable to those of leading companies like OpenAI and Anthropic, but it was seemingly created with a fraction of the resources.
Throughout Monday morning, the app experienced outages, which it said were from high traffic. And it temporarily limited registrations due to a cyber attack. Still, DeepSeek quickly became the most downloaded free app on Apple’s app store, overtaking ChatGPT.
I spent the morning playing with the chatbot, asking it, along with OpenAI’s ChatGPT and Anthropic’s Claude, all the questions I could think of. After some initial toying, I was impressed.
It was able to solve some complex math, physics and reasoning problems I fed it twice as fast as OpenAI’s ChatGPT. When I asked it questions about computer programming, the types a job applicant might be faced with in a technical interview, its responses were as in-depth and speedy as its competitors.
When I gave DeepSeek prompts that required it to scrape the web for answers, like to write biographies for some of my co-workers, DeepSeek appeared to have fewer hallucinations than ChatGPT, though its answers felt slightly worse when composing poems and short stories, planning vacations and coming up with dinner recipes.
The model had other weaknesses. DeepSeek was heavily censored for American users. When I asked it to summarize the 1989 Tiananmen Square massacre, an event that the Chinese government has long tried to erase from the internet, it responded that the information was “beyond my current scope.”
