Assuming, as it seems, that this is a good, well-designed randomized experiment such that contaminating variables between group gender ratios and collective intelligence are scrambled, it's still worth keeping in mind that correlation is not causation.

So, completely apolitically, this seems to be an initial result that's only been replicated twice. It still needs to be borne out as to whether it holds more generally. The researchers seem to have a pretty plausible theory to explain the results, though, and further testing should show more insight.

But secretly I'm going to hold back a smallish bet that this'll be one of those "disappearing" effects that social scientists have been complaining about recently.