It is, though. Charles Stross addresses this and points out that you need a robot capable of human-level cognitive ability and at least the ability to recognize and simulate emotions if you want to raise functional human beings, because the robots are going to be the sole care-givers. So human-equivalent AI, in other words. Which renders sending humans superfluous. (Especially given the issue of how humans would survive in a hostile alien environment.)
Yeah, one of the existential threats is a gamma-ray burst from a nearby star going supernova. This would be equally deadly on Mars or even on a nearby, i.e. reachable, exoplanet. Less so at the bottom of the ocean.
Life on any other planet (or station or spacecraft) is necessarily going to be inherently more precarious than on the Earth we’ve evolved to live on, so even as a long-term survival strategy, it’s not great. We’d colonize other planets and then have those colonies be wiped out multiple times over before the human population on Earth faced similar existential threats.
