I'm fairly certain it's humans that will destroy Humanity.
This is doom porn. Algorithmic agents operating on large data sets are very useful and there zero dangers of such agents taking over the world to make paperclips. The more immediate danger is corporations (which are a form of AI) and their short-term profit objectives.
Yeah, it's nerd-sniping, distracting them from actual global problems like energy poverty, malnutrition, endemic disease, ecosystem collapses and tragedy of the commons problems, pollution and cumulative poisoning, oppression of women. And so on.
My point is that these articles are basically useless and people should stop writing them. Talking about AI requires an understanding of formal systems and it's clear that most authors lack such understanding.