Athira Sethu
Kochi, 8 February 2025
DeepSeek, a Chinese start-up, stunned the world on January 27, 2025, when it released an AI model that was as good as the tech giants like OpenAI, Meta, and Google. DeepSeek’s AI assistant topped the most downloaded app in the U.S. overnight, shocking many who believed only the big Western companies could dominate AI. The company’s model was developed at a much lower cost compared to models like ChatGPT and Gemini, shaking up the market and being called “AI’s Sputnik moment.”
The novelty in DeepSeek technology is the efficiency of the optimized software and hardware instead of costly chips. The other models turn on all the parameters at the same time. In contrast, DeepSeek’s model employs the Mixture-of-Experts approach, where it activates only the most relevant parts of the AI. This helps in minimizing computing needs. In addition, they used cheaper Nvidia H800 GPUs and PTX for performance optimization. Thus, they could construct powerful models without incurring heavy costs.
Established late in 2023, DeepSeek moved with lightning speed to shake up the AI world; it released its model within the following year. Even when advanced Nvidia chips became restricted by the U.S., DeepSeek optimised its model to run on more commonly accessible hardware.
DeepSeek’s models are open-source, meaning anyone can access and modify them without any restrictions. Unlike companies such as OpenAI and Google, DeepSeek’s approach does not limit access to their AI models. This is where DeepSeek allows innovation through open-source models that independent developers and researchers can further improve the technology. This is a good strategy, but the potential for misuse is huge in the creation of deepfakes or misinformation.
Launched into the global community, this makes advanced AI even more widely accessible, maybe bringing about improvements more quickly within realms such as medicine and the sciences. For every positive spinoff, risks from security or potential misuse due to lack of control over distribution, as such there is not any central monitoring to regulate distribution or use remain with DeepSeek. Some states such as in the U.S, Taiwan and Italy have taken actions to freeze access to use due to concern with national security.
This calls for regulation because the technology develops at a much faster pace than laws can follow. Governments should come up with ways to foster innovation while, at the same time, seeing that AI is developed and applied responsibly. The DeepSeek model dislodges the hegemony of large firms and opens the avenue of developing AI to developers, but in this regard, it is very challenging to control and manage possible risks.
The future of AI development and impact will depend on how open AI models should be and who controls them. The debate over these issues will shape the next phase of AI technology.
