I'm not sure these are even actionable enough to serve as "general principles".

Take Walmart, for example. Its massive global supply chain channels money to countries that otherwise wouldn't have as much to offer economically, on balance decreasing inequality. By exploiting its economy of scale to warp the domestic U.S. market, it is able to pay workers--already at the low end of the income distribution--less than at competing stores (e.g. small collection of local stores), increasing inequality. But the comparatively inexpensive goods allow a wider segment of the U.S. population to have convenient and economical access to pretty decent stuff, decreasing functional inequality. The Waltons (and wealthy investors) profit handsomely, increasing inequality.

If AI could do a Walmart-equivalent, then we would say...what?

I think that principles, such as they are, will have to be substantially less airy than this. Asimov was, I think, more on track with the style with his Laws of Robotics: very simple, very immediately applicable. And even that was enough to still admit a pretty wicked novel.