I think the bias you're criticizing is teleology. [1] I suspect we have something akin to a Purpose Co-Processor. Further that I'm using said co-processor to annotate it as such. The inclination to trust these annotations is intuitive and automatic. To be valuable they only need to be optimal and not necessarily true. Resisting this urge should be part of the instruction for any epistemic discipline. Like evolutionary theory. The problem with most "people that are otherwise intelligent" is that they're optimality junkies.

1: https://en.wikipedia.org/wiki/Teleology