After the huge attention gained by deepseek R1, I have seen a good amount of people immediately went to looking for censorships (which are expected to exist), laughed at it, and looked down on it. They totally missed the points and these behaviors made them look like slaves of their very own defense mechanism1.
Deepseek did make some techonological breakthroughs. Indeed the censorship is a defect of such a great product. However making jokes on it with its censorships is just like catching grammatical errors in an insightful article and disdain the article because of that. Yes, grammatical errors are errors, catching itself could be useful, but doing that for attacking an insightsul article is another thing. It just misses the point.
If you see China as a rival, it is also like laughing at your enemy’s weapons just because they “look” funny.
If you want to disdain an insightful article, disdain by its ideas.
The accuses about cheating with OpenAI data are even more related and look less like jokes. (if ignore the validity of those accuses)
An even more related critiques can be pointing out the fact that it did not push the limit of AI reasoning ability (since it is roughly as good as o1), although it optimized the whole process. Of cause making training cheaper is important and it helps researchers to push the limits of AI as well (since the testing cost became lower as well). Depends on your interested areas, you could value Deepseek differently. <- This, is still way better than taunting it because of the censorships.
Since they could not bare the fact that Chinese make breakthroughs that others did not make. This is called “glass-made heart” in Chinese, ironically mostly used in describing the hysterical behaviors of pro-CCP people when they heard something that is unfavorable to CCP. ↩︎