I don’t think that’s true. There are lots of questions humans deal with that have the following properties:
Typically, they are questions about human affairs rather than the hard sciences.
We answer them, far from perfectly, but still much better than any computer program yet devised.
We do this at least in part by integrating large volumes of weak evidence.
It seems likely—I would go so far as to say beyond reasonable doubt—that no algorithm that tried to instead use precise inference on smaller volumes of evidence could match human performance on such questions, given realistic constraints on available data and thinking time.
I don’t think that’s true. There are lots of questions humans deal with that have the following properties:
Typically, they are questions about human affairs rather than the hard sciences.
We answer them, far from perfectly, but still much better than any computer program yet devised.
We do this at least in part by integrating large volumes of weak evidence.
It seems likely—I would go so far as to say beyond reasonable doubt—that no algorithm that tried to instead use precise inference on smaller volumes of evidence could match human performance on such questions, given realistic constraints on available data and thinking time.