Chinese start-up DeepSeek has caused quite a stir in the AI market in recent days. Their free chatbot, which is similar to ChatGPT, is currently one of the most popular downloads worldwide. What makes DeepSeek special is that their AI model is trained on cheaper chips, which has cut costs by tens of millions compared to established names such as OpenAI.
Even Sam Altman, CEO of OpenAI, spoke highly of DeepSeek's new model. On X, he called it an "impressive model, especially considering what they've accomplished for this price." Still, Altman seems cautious about the idea that AI development will become cheaper in the future. According to him, "greater computing power remains essential to achieve our mission."
The use of data and censorship
By using cheaper chips, DeepSeek can process large amounts of data more efficiently and train the model faster. The model is available open source. It does raise questions about other costs, data security, as well as potential security risks. How is all this data stored, who has access to it, and how vulnerable is this information to attack or misuse? The lack of transparency about these processes increases the risk of sensitive data falling into the wrong hands.
Therefore, it is essential to think carefully about what information you share with AI platforms such as DeepSeek. This risk becomes especially clear now that it appears the chatbot is currently avoiding certain sensitive topics, such as freedom of speech or human rights violations - raising questions about ethical and societal implications.
Accelerating AI developments
With the introduction of DeepSeek, we expect developments in the AI sector to be even faster. At the same time, managing these advances remains a challenge. We are following the market closely and eagerly awaiting what the future holds.
Learn more about the innovative technologies behind DeepSeek V3 in the article by our AI expert Furkan Karacaer .