The emergence of DeepSeek's R1 chatbot is challenging American dominance in AI, raising concerns among US tech experts and investors about the implications for the industry.

This development reflects not only the current hype surrounding AI but also the geopolitical tensions, particularly given its origins in China.

DeepSeek's chatbot was unveiled at the end of January 2025, coinciding with OpenAI's announcement of a new reasoning model, o3-Mini, in response to the competitive landscape.

Following DeepSeek's announcement, shares of major tech companies, including Nvidia, experienced a significant drop, erasing nearly AU$1 trillion from its market value.

DeepSeek's R1 model utilizes a reasoning model and is designed to avoid sensitive topics like Chinese politics, reflecting its training data and cultural context.

DeepSeek, a Chinese startup founded in 2023 and led by mathematician Liang Wenfeng, has launched its R1 chatbot, which is gaining global attention for its competitive capabilities against leading American models.

Despite its impressive features, DeepSeek has been criticized for providing false claims 30% of the time and failing to answer 53% of questions, indicating a high unreliability rate compared to its competitors.

DeepSeek's innovations could democratize AI development, allowing more companies to train their own models and reducing reliance on major tech firms.

The competition between US and Chinese AI models highlights a broader battle of values, with significant implications for global soft power dynamics.

US technology secretary Peter Kyle expressed both excitement over DeepSeek's breakthrough and concern regarding the data privacy and security risks associated with the Chinese app.

The US ban on advanced chip sales to China may have inadvertently spurred innovation among Chinese researchers, contributing to projects like DeepSeek.

While DeepSeek does not technically advance beyond existing large language models, it shares similar flaws, such as hallucinations, raising questions about its overall effectiveness.