Messing around with more ChatGPT coding experiments, and it’s consistently failing basic SWE-interview starter questions. As I correct with counter-examples to the generated code, ChatGPT seems to modify its understanding of the problems to match the bad code, not the reverse
Wired Magazine, 2006: “Silicon Valley is roaring back to life as startups mint millionaires and Web dreams take flight. But no, this is not another bubble.”
I actually think this article was fantastically on the money – that cheap bandwidth, open source software, and (what they’d didn’t yet call) cloud services would lead to an explosion of low-cost startups with big upside
