The chip ran Linux benchmarks at about i486 speeds.
The speed is not impressive, but the fact that did it at all is interesting.
Is it, though?
It’s a clearly defined spec, a clearly defined utility function and a very limited set of building blocks. It’s not trivial, but not super impressive either.
And given the nature of this kind of research and the trackrecord of Chinese scientists in particular, I’d assume it’s getting less and less impressive the deeper you look.
What kind of long term things can happen?
There a story about AI learning some industry secrets / trade secrets
https://www.techradar.com/news/samsung-workers-leaked-company-secrets-by-using-chatgpt
I wonder if this would end up enabling sanctioned states to step around the chip import ban or nuclear weapon constructions.
