Strange the way anti-hype works just like hype. From the results I've seen,
DeepSeek's answers are far worse than, say ChatGPTs. They're much vaguer and more rambling, often largely unrelated to the question asked. Not only do they read like an undergrad essay, they read like the student was drunk.
Also, when there are export restrictions on stuff you're buying from another country and you report that you came in just under the magic number, well, surprise, surprise.
Sure, there's no reason Chinese researchers couldn't do this stuff just as well as the US or better. And, yeah, the AI bubble bursting is a real popcorn moment for us all, all the same.