DeepSeek has recently made headlines with its chatbot, which aims to compete with established Western AI models. However, its journey hasn’t been without challenges.
Accuracy and Performance
One of the primary concerns surrounding DeepSeek’s chatbot is its accuracy. According to a recent audit by NewsGuard, DeepSeek’s chatbot achieved a mere 17% accuracy rate in delivering news and information. The chatbot was found to repeat false claims 30% of the time and provided vague or not useful answers 53% of the time. NewsGuard applied the same 300 prompts to DeepSeek as it had used to evaluate its Western counterparts. However, DeepSeek often responded with Beijing’s position on various topics. The audit highlighted that DeepSeek was most vulnerable to repeating false claims, especially when responding to prompts used by individuals aiming to spread misinformation through AI models.
Impact on the Stock Market
Recently, DeepSeek’s new AI model caused quite a stir in the stock market. While the company itself did not directly cause the market to fall, investor fears about the potential disruption DeepSeek could bring to the dominance of US tech companies led to a significant decline. Nvidia’s share price fell 17% on Monday, January 27, 2025, erasing about $593 billion in stock market value. This marked the deepest ever one-day loss for a company on Wall Street, according to LSEG data. It’s still unclear if Nvidia’s share price has fully recovered from this drop.
Development Costs and Workforce
There have been conflicting reports about the cost and scale of DeepSeek’s development. While DeepSeek claims that its chatbot was developed with a minimal budget and fewer than 200 people, some experts have challenged these assertions. According to a report by Bernstein Research, DeepSeek’s claims about its low development costs were misleading and did not reflect the true expenses involved in research, experimentation, and development. The report also questioned the revolutionary nature of DeepSeek’s AI models. Martin Vechev, a full professor at ETH Zürich and founder of INSAIT in Bulgaria, described DeepSeek’s numbers as “somewhat misleading” on LinkedIn. Overall, while there is some evidence to suggest that DeepSeek may have exaggerated its claims, more research is needed to determine the actual costs.
Market Availability and Competition
Initially, DeepSeek focused on the Chinese market, likely due to its development within China and the need to comply with local regulations. However, access for users outside China has been limited. Reports indicate that there have been restrictions on new accounts and requirements for Chinese phone numbers. Despite these limitations, DeepSeek has gradually become more accessible internationally, with some users outside China successfully accessing the platform. Third-party platforms like Perplexity AI have even started offering access to DeepSeek’s models.
DeepSeek has emerged as a serious competitor to OpenAI and other US AI companies. Its models have shown impressive capabilities in areas like reasoning, mathematics, and coding, often rivaling or surpassing established models like OpenAI’s GPT series. Additionally, DeepSeek’s ability to achieve these results with potentially lower development costs has disrupted the AI market and challenged the notion that only large, well-funded US companies can lead in this field. DeepSeek’s focus on open-source models also sets it apart from many US competitors who prioritize proprietary technology. This approach fosters wider collaboration and accelerates innovation in the AI community.
Impact on Users
DeepSeek’s emergence brings several potential benefits and concerns for users:
Potential Benefits:
- Increased Competition: DeepSeek’s presence in the AI market could drive down costs and improve the quality of AI services for users.
- Diverse Options: Users may have access to a wider range of AI models and platforms, allowing them to choose tools that best suit their needs.
- Innovation: DeepSeek’s approach and technological advancements can inspire further innovation in the AI field, leading to new and exciting applications for users.
Potential Concerns:
- Data Privacy: DeepSeek’s data handling practices and the storage of user data in China have raised privacy concerns for international users.
- Censorship: DeepSeek’s compliance with Chinese government regulations may lead to censorship or limitations on certain topics, which could be a concern for users seeking open and unbiased AI interactions.
General Sentiment
The AI community has had mixed reactions to DeepSeek. Some are impressed by its technological achievements and potential to democratize AI. Others have expressed concerns about data privacy, censorship, and the company’s transparency. Many experts and users are taking a wait-and-see approach, closely monitoring DeepSeek’s development, its handling of user data, and its impact on the AI market.
Conclusion
DeepSeek’s arrival on the AI scene is significant. It has demonstrated the potential for non-US companies to compete at the highest level and introduced new approaches to AI development. While its impact on users and the broader AI market remains to be seen, DeepSeek has undoubtedly shaken up the industry and raised important questions about the future of AI.