[hulitu]

> You don't see evidence of bias there, when people with darker skin are an afterthought?

I see a bias. However, maybe the testing was done only on poor white people because that's what the testing company employed (due to operating in a poor country with no significant "coloured" people population). Everybody wants to save money. Test results almost do not matter if the test is "done". Especially when "best practice" comes from SW.

[Jordrok]

This is the entire point of the article - that inadequate testing can lead to inaccurate results. The point isn't that "the device is biased and therefore its makers are bad people", it's to highlight a clear-cut case where cutting costs by doing such limited testing leads to negative outcomes.