Maybe I'm being slightly too cynical, but that's pretty much inevitable. A great deal more "scientific knowledge" than you'd probably be entirely comfortable with knowing involves taking a point sample of some space and applying it way, way away from where the sample was actually taken. If this study succeeds, then you can expect the HN commentariat to be citing how BI has been scientifically proved to be a good idea until the end of time, and if this study fails, the opposite.
Even better, there's really two types of BI going on here. One is as a mechanism for poverty alleviation, by getting capital to people who have none. Not just "not much", but none. The other is as a solution to the as-yet mostly future problem of the work force being automated out of existence in every industry that people below a certain IQ threshold can work productively in. Expect those to be conflated freely, even though in my opinion they're two different problems. (If you squint really hard, yes, they blur together, but don't ignore the fact that you're squinting.)