by DeepSeek R1, a Chinese chatbot that competes with ChatGPT, has gained significant attention due to its low price and rapid app downloads. The chatbot’s launch caused a significant financial market reaction, with tech stocks like Nvidia experiencing a 17% plunge in a single day.
DeepSeek’s entry sparked debate about the significant financial investment required to build artificial intelligence. The story raised questions about the justification for the $500 billion Stargate investment, which was revealed by Donald Trump. However, deeper investigation revealed that DeepSeek was actually constructed for only $6 million, revealing more than initially believed. This tech story has significant implications for the current AI revolution.
What is DeepSeek exactly?
DeepSeek, a Chinese natural language chatbot, uses an LLM similar to ChatGPT. Despite being built for only $6 million, it has comparable power to ChatGPT. Despite OpenAI’s billions in funding, it has not disclosed the exact cost of developing its own chatbot.
Similar to ChatGPT, DeepSeek can generate almost anything that can be written by a person, including computer code and poetry, given a prompt. It is able to accomplish this because, like all large language models (LLMs), it has been trained on vast volumes of text—the costly aspect of creating artificial intelligence.
However, DeepSeek’s developers have made it open-source, which sets it apart from ChatGPT. This implies that anyone can see, use, or alter the code however they see fit. In contrast, ChatGPT is a proprietary program, and the specifics of how it operates are still under strict confidentiality. Curiously, this open feature provides us with a hint as to how it was made so economically.
Meta’s chief scientist, Yann LeCun, revealed that DeepSeek is built on pre-existing AI code, including Meta’s Llama LLM models, using an open-source methodology. This could potentially lower the training cost, but the release of DeepSeek led to the largest decline in IT stock value in recent years.
Is AI Too Expensive?
Concerns have arisen about the justification for large investments in AI businesses, particularly in the US, which could significantly impact the global economy. Tech stocks have contributed to the rise of US stock markets over the past decade, believing their AI investments will yield substantial returns.
DeepSeek’s achievements are attributed to its lack of access to the latest Nvidia AI processing technology, which is subject to a U.S. government ban on exports to China, resulting in DeepSeek relying on performance-capped versions of approved chips.
The sanctions on China’s AI may have acted as a catalyst for innovation, as startups like DeepSeek prioritize efficiency, resource pooling, and collaboration, rather than weakening their AI capabilities.
The Question of Privacy
DeepSeek, despite its popularity, raises privacy concerns due to its data sharing with its Chinese parent company, despite not disclosing this, preventing legal issues or potential GDPR breaches.
Chinese chatbots, like ChatGPT and Google’s Gemini, require users to consent to data sharing. However, unlike US companies, Chinese companies are free or obliged to share collected information with their government, potentially allowing the Chinese government to access vast amounts of Western citizens’ personal data.
DeepSeek, a popular AI platform, was targeted by a massive cyberattack shortly after its launch, allegedly a distributed denial of service (DDoS) attack, aiming to take the service offline. This incident is similar to the TikTok controversy, as it raises concerns about potential advantages for AI system training.
A New Age of AI?
DeepSeek’s successful launch is a triumph of open source over proprietary AI development methods, as it fosters innovation and promotes sharing of ideas and information. LeCun highlights the power of open research and open source, as everyone can profit from their published work, highlighting the potential of open research and open source.
Donald Trump, after announcing a half-billion investment, expressed optimism about the potential of Chinese companies to create powerful AI tools at a fraction of expected costs, suggesting a shift towards a more democratized development landscape.
Open source and its community of collaborative, innovative builders will play a crucial role in the advancement of cutting-edge technology, despite the dominance of giants like Microsoft and Google. If geopolitical issues are resolved, this will benefit everyone.
