I think the causality on slavery is backwards, and the article agrees... Shortages of low-skilled labor encourage automation, because then the (very expensive) automation process becomes cheap relative to labor. The article argues that labor shortages coinciding with excess capital paved the way for automation, which in turn reduced dependence on slavery in the long term.

In this sense, cheap labor is cheap, but also a bit of a trap: it doesn't scale into undreamt of use-cases the way that truly scalable automation tends to.