It's important to separate the marketing term from the underlying collection of machine learning technologies. In 2003 I was having chatbot experiences that felt identical to modern ones. In the 2010s I was spending hours trying to train speech-to-text algorithms for medical notation. Computer vision started maturing in the 1990s. Most of what we consider AI today is just a lipstick pig made out of things that weren't sexy a decade before that became an economic sector. The things the AI industry brought on top of that will go away while those 2000s technologies still get refined.

I don't think the Chinese AI industry is better for them. I just think it's worse for the US because it has run out of tech bubble ideas and that's the only thing that has kept it alive since the 1990s. There's nothing to carry the US economy after AI beyond healthcare and tourism that nobody can afford. Deepseek is just as useless to a Chinese guy as it is to me, but it does ChatGPT better than ChatGPT no matter how much the US autocannibalises to fund OpenAI. That just adds extra layers to my criticism which is as a neo-Luddite. It kills the thing I hate faster by making it more visibly useless.