DeepSeek: A Wildly Cheap AI Model with "Chinese Characteristics"
Lots of recent news about the DeepSeek R1 models, from Chinese AI lab DeepSeek. It is a large language model (LLM) with long reasoning chains that can generate text, translate languages, and answer questions. DeepSeek is trained on a massive dataset of text and code, and it is able to perform a wide range of tasks. However, DeepSeek is also explicitly coupled with CCP propaganda. For example, when asked about Taiwan, DeepSeek responds that it is a part of China. When asked about Tibet, DeepSeek responds that it is an autonomous region of China. And when asked about Tiananmen, DeepSeek responds that it was a "counter-revolutionary riot."
Similarly, China's cheap manufacturing has been coupled with dealing with or accepting the CCP's views on politics for decades. Western companies have been able to source products from China at very low costs, but this has often come at the expense of human rights and environmental standards. In some cases, Western companies have even been complicit in forced labor and other abuses. This rhymes with how DeepSeek is a powerful AI model that is available at a very low cost, but it comes with the CCP's propaganda.
Every maker of LLMs embeds their own political and cultural views into them when they ready them for release to the world. As currently made, you are never purchasing just an AI model. You are purchasing an AI model with the ethics/politics of its creator. This is something everyone needs to be aware of when they are considering using any AI Model.
If you want an overwrought further exploration of this, check out my post on how this happens.