10. Triple-E Framework to Evaluate The future of DeepSeek

DeepSeek is one of China's more promising AI research institutions, particularly demonstrating significant progress in developing large language models (LLMs) and AI reasoning. Its long-term development potential can be analyzed from several aspects:

1         

2         

3         

4         

5         

6         

7         

8         

9         

10         

10.1        Technical Strength and Development Potential

• Model Capability Improvement: DeepSeek has released the DeepSeek LLM and is continuously improving it. It may further develop architectures such as DeepSeek-MoE (Mixture of Experts) to enhance computational efficiency and reasoning capabilities.

• AI + Quantum Computing: If China achieves breakthroughs in quantum computing, DeepSeek may integrate quantum computing technology to strengthen AI training, which could be a key advantage in future AI competition.

10.2        Industry and Commercial Applications

• Enterprise Application Expansion: If DeepSeek can integrate with major Chinese enterprises (such as Alibaba, Tencent, and Huawei) and government institutions, it could exert significant influence in AI industry standards and enterprise intelligence applications.

• Open-Source Strategy and Global Competition: If DeepSeek adopts a more aggressive open-source strategy (similar to Meta Llama), it could attract more developers, increase its global influence, and even compete with OpenAI and Anthropic.

10.3        International Politics and Policy Impact

• China’s AI Independence: Due to U.S. sanctions on high-end chips, China’s AI development is moving towards "de-Americanization." If DeepSeek can efficiently run AI models on domestic hardware (such as Huawei Ascend chips and Tianhe supercomputers), it will become a key player in China’s AI industry independence.

• Global AI Standard-Setting Power: If China can promote AI safety and ethical standards, DeepSeek may participate in the formulation of international AI standards, challenging OpenAI and U.S.-led technological norms.

10.4        Competition and Challenges

• Talent and Innovation Environment: The Chinese AI industry experiences high talent mobility. DeepSeek’s ability to attract and retain top AI researchers will be crucial for its long-term success.

• Capital and Market Competition: Compared to competitors like OpenAI, Google DeepMind, and Anthropic, DeepSeek still needs more financial support and a larger market share.

10.5        Summary 1

DeepSeek has strong development potential in terms of technical strength, industrial applications, and policy support, particularly in China’s AI independence and international standard competition. However, it must overcome hardware limitations, talent competition, and global market expansion challenges. If it successfully integrates with China’s AI ecosystem and builds global influence, DeepSeek could become a core engine of China’s AI industry.

10.6        DeepSeek and International Data Privacy Issues

The United States said that Deepseek's software contained a program that transmitted user data to China Mobile, allowing the Chinese government to obtain user information.  DeepSeek, as a Chinese AI company, is indeed considered a potential data security risk by the U.S. government due to its technological background and China's regulatory environment. The U.S. has recently intensified its scrutiny of Chinese tech firms, particularly in AI and data privacy. If DeepSeek is accused of transmitting user data back to China, it may face regulatory challenges in the U.S. similar to those encountered by TikTok and Huawei.

However, whether DeepSeek actually transfers user data to China Mobile or the Chinese government requires concrete evidence. If DeepSeek intends to expand internationally, it must strengthen its data privacy policies and may need to establish independent data centers overseas (similar to TikTok’s “Project Texas” in the U.S.) to alleviate concerns about data compliance from various governments.

Potential Impacts of U.S. Allegations

1. Increased Restrictions on Chinese AI Companies: If the U.S. government deems DeepSeek a national security risk, it may ban American companies from collaborating with DeepSeek, restrict its market presence in the U.S., and even impact partnerships with U.S. semiconductor firms like NVIDIA and AMD.

2. Global Ripple Effect: If the U.S. places DeepSeek on an entity list, countries such as the EU, UK, and Japan may follow suit, further limiting DeepSeek’s competitiveness in the global market.

3. Response from the Chinese Government: In response, the Chinese government may provide more support to DeepSeek to ensure its technological development remains unaffected. Additionally, China could intensify scrutiny of U.S. AI firms, potentially restricting products from companies like OpenAI and Google AI in retaliation.

Overall, if DeepSeek can enhance transparency in its data processing practices and comply with international privacy standards (such as GDPR), it may mitigate the impact of such allegations on its global expansion. However, if U.S. sanctions escalate, DeepSeek may be forced to shift its focus toward China and other friendly markets, such as the Middle East, Southeast Asia, and Africa.

Insightful Discussion of DeepSeek LLM

表單的頂端

If anyone downloads DeepSeek's open source code, makes sure there is no function in the code to transmit user information to the Chinese government, and uses very cheap hardware to host DeepSeek LLM for everyone to use, then it will be impossible to prevent DeepSeek's software from being accepted by the world.  If DeepSeek’s open-source code undergoes scrutiny by the international community and is confirmed to have no hidden data transmission mechanisms, and if anyone can self-host DeepSeek LLM on their local machine (or on affordable hardware), then the effectiveness of U.S. or other governmental bans will be greatly diminished.  Let us analyze this problem:

10.7        Why Can’t DeepSeek Be Stopped?

1. Transparency of Open-Source Code

   • If DeepSeek’s LLM is truly fully open-source, anyone can inspect its code to ensure it does not secretly transmit data back to China or elsewhere.

   • The global developer community (such as those on GitHub and Hugging Face) can maintain and improve the model collectively, preventing any single government from monopolizing it.

2. Local Deployment vs. Cloud-Based Restrictions

   • The U.S. government can block DeepSeek’s official API, but if DeepSeek LLM can be self-hosted, then anyone with a sufficiently powerful computer (or even a Raspberry Pi + small GPU) can use it without relying on DeepSeek’s official servers.

   • This decentralized hosting model makes government restrictions much harder to enforce, as DeepSeek LLM would not be concentrated on a specific server but could instead operate independently worldwide, similar to BitTorrent.

3. Support from Open-Source Communities

   • Open-source AI communities (such as Hugging Face, LAION, and EleutherAI) might provide technical support for DeepSeek, helping it optimize, enhance, and adapt to more hardware.

   • If DeepSeek LLM is powerful and free, a large number of developers will adopt it, further solidifying its influence.

4. U.S. Government’s Tech Ban Cannot Directly Affect Individual Use

   • The U.S. government can prohibit OpenAI, Google AI, or Anthropic from collaborating with DeepSeek, and even prevent U.S. companies from using it, but individual users can still download and run it locally, just like how Midjourney cannot block Stable Diffusion.

   • As long as one person downloads it from GitHub, the model can never be completely banned.

10.8        Potential Challenges for DeepSeek LLM

Despite these advantages, DeepSeek may still face technical and policy-related challenges:

1. Hardware Supply Constraints

   • If the U.S. further restricts AI GPU exports (such as NVIDIA A100/H100), DeepSeek’s ability to train more advanced models may be affected, making it harder to keep up with OpenAI or Google.

   • However, open-source LLMs do not require DeepSeek itself to continuously train new models—if the current DeepSeek LLM is good enough, the community can continue improving it.

2. Trust Issues in International Markets

   • Even if DeepSeek’s open-source model is clean, some companies or governments might hesitate to use Chinese-developed technology for political reasons, opting instead for U.S. or European LLMs.

   • However, this will have little impact on individual developers or the free market, as users tend to choose the cheapest and most powerful tool, regardless of its origin.

3. Potential U.S. Pressure on Open-Source Communities

   • The U.S. government may try to ban U.S. companies from hosting DeepSeek models (for example, by forcing GitHub or Hugging Face to take down DeepSeek LLM), but this would not prevent other countries from hosting these models.

   • Open-source communities in China or Europe might take over the hosting and maintenance of DeepSeek, making bans ineffective.

10.9        Summary 2

If DeepSeek LLM is truly open-source and can be run locally, then the U.S. government will find it nearly impossible to stop its spread. As long as developers continue using it, DeepSeek can keep evolving through the open-source community and could even become one of the mainstream open-source AI models in the future.

The real key question is: Is DeepSeek genuinely fully open-source? If it only offers an open API but not the full model, then the U.S. government could still influence its growth through internet restrictions. But if DeepSeek releases the complete open-source LLM weights and training code, its impact will be unstoppable.