William F. Buckley was very intelligent, but so was Gore Vidal.
There are many, many topics where increasing intelligence does not mean increasing convergence of conclusions. The further away you get from hard science, the easier it is to diverge... building (intelligently) upon incorrect and hard-to-falsify foundations.
Somewhat predictive.
> The DOD considers it real enough that all recruits must pass a cogntive screening test, the AFQT, which is like a proxy for measuring IQ.
The DOD also often requires polygraph tests, so not the best endorsement.
IQ is a predictor of success in almost any particular metric for which it's reasonable to measure (so things like success in visual art and music don't correlate as well. ) In finances, career paths, academia, social status, politics, writing and so on, IQ strongly correlates to relative performance.
IQ may be a poor proxy for some strictly well derived fundamental of cognition, but we don't have that quite yet.
Maybe we'll be able to ask gpt-9000 and get a sensible answer?