What’s the story on DeepSeek?
Imagine AI being like the newest smartphone – everyone wants it, but those who make it have been spending big bucks on the chips, screens and other tech that goes into the manufacturing of it. DeepSeek comes along and shows you can get a phone that’s almost as good, but for only 10% of cost of an iPhone or latest Android. Now, everyone’s wondering if they’ve been overpaying for all those fancy parts.
Why the market freak-out? Well, companies like Nvidia have been like the kings of the chip world, with their chips being the gold standard for AI. DeepSeek’s approach means you might not need those expensive chips to make cool AI stuff. This has investors worried because if AI can be done cheaper, maybe they’ve been betting on the wrong horse all along.
So, the market got scared, thinking “If AI can be this good and this cheap, what does that mean for all the money we’ve poured into the high-end stuff?” That’s why you saw stocks take a dive – it’s like finding out you can buy a car equivalent to a new Mustang GT for $5,000 versus $60,000.
Of course, the other bad news is it is Chinese.
According to Wired:
To be clear, DeepSeek is sending your data to China. The English-language DeepSeek privacy policy, which lays out how the company handles user data, is unequivocal: “We store the information we collect in secure servers located in the People’s Republic of China.”
It has also been pointed out that DeepSeek will not respond to subjects that reflect negatively on China. This is similar to the way Google Gemini and ChatGPT operated when it came to conservative politics. That is why I went to X and got Grok.
If you thought TikTok was going to be controversial, guess again.
Will DeepSeek give China a military advantage?
DeepSeek could potentially give China a military advantage by enhancing AI applications in areas like autonomous systems, strategic analysis, and cyber operations. However, realizing this advantage depends on successful integration, counteractions by other nations, and broader strategic contexts.
The U.S. has the technological, policy, and strategic capabilities to counter DeepSeek through innovation, alliances, and regulatory measures. However, success depends on adaptability and timely response to advancements.
U.S. policy, particularly export controls, played a role in shaping how DeepSeek was developed but was one among many factors contributing to its creation. China’s AI development is also the result of its own strategic planning, significant domestic investment, and a push towards technological self-sufficiency.
Can we trust what the Chinese claim?
I don’t know. However, I do know Musk is skeptical.
Elon Musk has expressed skepticism regarding the claims made by DeepSeek about their AI development, particularly concerning the number of Nvidia chips used and the cost-effectiveness of their model. Here’s how this skepticism has been portrayed:
- Skepticism on Chip Usage: Musk has indicated that he believes DeepSeek might have used far more Nvidia chips than they publicly admit. In response to a claim that DeepSeek only needed 2,000 Nvidia chips to train their model, Musk commented with a skeptical “obviously,” suggesting that the actual number might be much higher, possibly up to 50,000 Nvidia H100s, as speculated by Scale AI CEO Alexandr Wang. This implies Musk thinks DeepSeek’s achievements might not be as cost-effective or groundbreaking as claimed due to hidden resource usage.
- Response to Efficiency Claims: When Salesforce CEO Marc Benioff suggested that DeepSeek’s success proves AI isn’t just about GPUs, Musk responded with “Lmao no,” indicating his disbelief in the narrative that DeepSeek achieved its results without significant hardware investment.
- Market Impact: Musk’s skepticism likely contributes to the market’s reaction, as his comments can influence investor perceptions and market dynamics, especially given his standing in the tech and AI community.
From the information available, it appears Elon Musk indeed thinks that DeepSeek’s portrayal of their technology and efficiency might be exaggerated or not fully transparent. His comments suggest a disbelief in the simplicity and cost-effectiveness of DeepSeek’s approach to AI development.
Sources:
Here is what Grok says about why deepseek upsetting the market.
DeepSeek has caused significant disturbance in the market primarily due to its advancements in AI technology, offering models that are both competitive and cost-effective compared to established U.S. companies. Here are the key reasons based on recent information:
Cost-Effectiveness and Performance: DeepSeek’s AI models, like R1, have shown performance comparable to leading U.S. models such as those from OpenAI, but at a fraction of the cost. For example, DeepSeek-R1 is reported to be 20 to 50 times cheaper to use than OpenAI’s o1 model for certain tasks. This efficiency has raised concerns about the massive expenditures by U.S. tech companies on AI infrastructure, particularly on high-end chips like those from Nvidia.
Market Perception and Fear: The introduction of DeepSeek’s technology has led to a reevaluation of investment in AI technologies, especially in the stock market. The fear is that if AI can be developed at a lower cost, the market might have been overvaluing the need for expensive hardware and infrastructure. This fear contributed to a sharp decline in tech stocks, with Nvidia losing a significant amount of its market value on the day of the news.
Geopolitical Implications: There’s also a geopolitical angle where the U.S. has been trying to maintain a lead in AI technology, particularly through export controls on advanced chips to China. DeepSeek’s success with less advanced or restricted technology challenges this narrative, suggesting potential vulnerabilities in U.S. tech dominance and raising questions about the effectiveness of current export controls.
Investor Sentiment: The rapid success and popularity of DeepSeek’s app in app stores globally have caught investors off-guard, leading to a sell-off in tech stocks. This reaction was partly driven by the notion that the future profitability of AI investments might not be as guaranteed as previously thought, prompting a rethinking of investment strategies.
Overall, DeepSeek’s emergence not only questions the economic model behind current AI development strategies but also highlights the competitive landscape of AI technology, stirring market volatility as investors recalibrate their expectations and investments.
What does open source mean when it comes to DeepSeek?
When referring to DeepSeek, “open source” means that the underlying code, models, or other components of the AI system are made publicly available for anyone to view, use, modify, and distribute. Here’s what it specifically entails for DeepSeek:
- Code Availability: The source code for the algorithms or models might be available, allowing developers to see how DeepSeek works internally.
- Model Weights: For AI models like DeepSeek-V3 and DeepSeek-R1, the model weights (the parameters learned by the model during training) are shared. This allows others to use these models directly, fine-tune them for specific applications, or study them to understand or improve upon the AI techniques used.
- Licensing:
- DeepSeek-V3 is available under a custom license agreement which might have specific terms for use, modification, and distribution, but it’s still considered open source in the sense that the model weights and possibly some code are accessible.
- DeepSeek-R1 is now under the MIT License, one of the most permissive open-source licenses, which allows for free use, modification, and distribution of the software with minimal restrictions, only requiring that the license and copyright notice are included in all copies or substantial portions of the software.
- Community Engagement: Being open source invites community contributions, which can lead to improvements, bug fixes, or new features being added to DeepSeek by people outside the original development team.
- Transparency and Trust: Open source AI models can foster greater trust as users and researchers can inspect the algorithms for biases, security, or ethical concerns.
- Innovation: It encourages innovation by allowing researchers and developers to build upon existing work without starting from scratch.
However, it’s important to note that while the models or certain parts of DeepSeek are open source, not all aspects of the project might be. For instance, training data, specific methodologies, or proprietary software components might still be kept private or under different licensing terms. Therefore, when using an open-source AI model like DeepSeek, you should always review the specific license terms to understand what you can and cannot do with the code or model.
Discover more from Veterans for Trump
Subscribe to get the latest posts sent to your email.