Earlier quoted context omitted.
Google in 2007: Do no evil. Google in 2017: Skynet didn't build itself people! We need more ML training data!
Skynet v. 0.1 didn't build itself. Clearly, though, once it becomes a minimum viable product, it will produce the next generations of itself. Or will it? If it's an AI intent on approximating human speech processing, why should it be any good at programming? See also https://www.fanfiction.net/s/9658524/1/Branches-on-the-Tree-... .
But ... ugh, rationalist fiction always makes me a little sad. I share the motivation, the annoyance at how pervasive the plot driven unrealistically stupid decision is in fiction. I want to like it, but it's always so saturated with authorial naivete, inexperience and smugness.
I guess I just wish they'd revere Iain M. Banks more and Yudkowsky less.