DeepSeek, a Chinese AI-chatbot app

DeepSeek, a Chinese AI-chatbot app launched last week, has sparked chaos in the US markets and raised questions about the future of America's AI dominance. Let's take a closer look at how the app works.

How Does DeepSeek Work?

DeepSeek functions like any other chatbot, resembling OpenAI's ChatGPT or Google's Gemini. Users open the app or website, ask questions, and receive detailed responses. However, DeepSeek is notably chatty and refrains from expressing opinions directly.

The chatbot often prefaces its responses with statements like "this topic is highly subjective," whether discussing politics (e.g., "Is Donald Trump a good US president?") or preferences (e.g., "Which is tastier, Pepsi or Coke?").

While it avoids direct comparisons, DeepSeek carefully weighs the pros and cons of itself versus ChatGPT, which responds similarlyKey Features and Limitations

Trained on data up to October 2023.
App version accesses real-time data; the website version does not.
Focuses on safeguarding against misinformation.
Handles heavy user traffic but can slow down under demand.

Unlike ChatGPT, DeepSeek censors itself on subjects banned in China. Sensitive topics like the 1989 Tiananmen Square protests are avoided, with responses often replaced by "let's talk about something else."

Market Impact and Technical Insights

DeepSeek has triggered discussions about resource-efficient AI development. It reportedly uses Nvidia chips (older versions) and is built on Meta's Llama architecture alongside Alibaba's Qwen.

According to Kayla Blomquist of the Oxford Internet Institute, DeepSeek might signal a shift toward less compute-intensive AI development, challenging US AI firms' monetization strategies.

"It is pointing to potential methods of model development that are much less compute and resource-intensive. We'll see what the next couple of months bring." – Kayla Blomquist

DeepSeek has similar usability and features to established US AI tools but operates under China's regulatory landscape. While it introduces cost-effective AI development techniques, its future impact remains uncertain.