I'd have to disagree with "they don't predict anything".
Do you think P(intelligent|good at brainteasers) > P(not intelligent|good at brainsteasers)?
I'm not saying it's the best test, most accurate, etc. but if someone is really good at brainteasers I'd post it's more likely than not that they're intelligent. There's probably false negatives with people who are intelligent but bad at brainteasers. But for a filtering test (when Google has too many applicants anyhow) I can see why they're used.
Do you think P(intelligent|good at brainteasers) > P(not intelligent|good at brainsteasers)?
I'm not saying it's the best test, most accurate, etc. but if someone is really good at brainteasers I'd post it's more likely than not that they're intelligent. There's probably false negatives with people who are intelligent but bad at brainteasers. But for a filtering test (when Google has too many applicants anyhow) I can see why they're used.