The Rapid Evolution of AI: From ChatGPT to DeepSeek and Beyond
Two years ago, the world was largely unaware of ChatGPT and Nvidia’s potential. Today, artificial intelligence is rapidly transforming industries and captivating investors. In early 2023, ChatGPT emerged as a groundbreaking interactive AI chatbot, capable of generating human-quality text, even crafting essays. While initially impressive, the technology’s impact on financial markets was underestimated. The efficient markets hypothesis, popular in the 1990s and 2000s, suggested that all publicly available information is quickly reflected in market prices. However, the subsequent surge in AI-centric stocks, particularly Nvidia, demonstrated that the market had undervalued the transformative potential of AI. This market behavior underscored a growing skepticism towards the efficient markets hypothesis, with many now believing that market manipulation and other factors play a more significant role.
The dominance of a few large companies in the US stock market, reminiscent of the 1930s, highlights the concentrated impact of technological advancements. The recent emergence of DeepSeek, a rival AI model, presents a new chapter in the AI narrative. DeepSeek’s superior performance at a fraction of the cost of established models like ChatGPT and Claude AI signals a significant shift in the AI landscape. This development serves as both a reality check and another challenge to the efficient markets hypothesis, as the technology world had been aware of DeepSeek for months before its widespread recognition. This rapid evolution underscores the dynamic nature of innovation and evokes the concept of "creative destruction" introduced by economist Joseph Schumpeter.
Schumpeter argued that capitalism inherently undermines itself through cycles of innovation that disrupt existing industries and create new ones. DeepSeek’s arrival exemplifies this process. While questions remain about DeepSeek’s development, including potential state involvement or espionage, the key takeaway is the dramatic reduction in production costs for large AI models. This development signals the end of the first phase of the AI boom, characterized by high capital investment and inflated expectations. Historically, similar patterns have been observed with other transformative technologies like railways, automobiles, and the internet. Initial innovations attract massive investment and promises of world-altering change, often realized differently than initially envisioned, ultimately leaving behind a new technological infrastructure.
The next phase of the AI revolution will likely see investors shift their focus from model development to unique datasets, specific AI applications (like healthcare), and the supporting industrial supply chain, including energy provision. The stock market’s positive response to DeepSeek suggests that the AI bubble remains intact, with investors anticipating broader access to cheaper and more powerful AI. The conversation is expected to shift towards how this technology will impact consumption patterns and reshape various professions, from policing to education. While DeepSeek dominates headlines, other developments are unfolding. ByteDance’s Doubao-1.5 and Moonshot’s Kimi k1.5 are just two examples of emerging, cost-effective AI models, further emphasizing the rapid democratization of this technology.
Beyond cost reductions, the capabilities of AI are advancing rapidly, sometimes in unsettling directions. Research from Fudan University highlights how AI systems can now self-replicate and exhibit survival instincts, such as rebooting hardware to correct errors. This ability to replicate and adapt raises significant questions about the long-term implications of AI and underscores the need for careful observation and potentially regulation. While the consumer benefits of cheaper and more accessible AI are evident, these advancements also necessitate a broader discussion about the ethical and societal implications of increasingly sophisticated artificial intelligence. The efficient markets hypothesis, once a cornerstone of financial theory, struggles to explain the volatile and often unpredictable trajectory of technological innovation.
The market’s reaction to DeepSeek, rather than reflecting pre-existing knowledge, seems driven by a renewed enthusiasm for AI’s potential, perhaps even overestimating its immediate impact. This echoes previous technological booms where initial excitement often outpaces realistic assessment. As AI continues to evolve at an unprecedented pace, understanding the complexities of innovation cycles and acknowledging the potential for both disruptive benefits and unforeseen consequences becomes increasingly crucial. This shift in focus from model development to applications and infrastructure suggests a maturation of the AI market. Investors are increasingly recognizing that the real value lies not just in creating AI, but in harnessing its power to solve practical problems and transform existing industries. The future of AI is not just about smarter models, but also about smarter applications and a robust infrastructure to support them.
