Oh really? A computer is nothing more than a collection of logic gates. You're essentially saying that humans are capable of building a super complex set of logic gates that have no errors. Are you sure? What about the FDIV bug? Exponentially complex machines have an increasingly higher chance of being flawed. Plus, until computers get infinite precision of decimals, a human with a pencil will always be able to do calculus more accurately.
I'm sure the logic gates did exactly what they were designed to do.
Unfortunately, they were following a flawed design.
The fun thing about science is that it can be tested, and refined. Our laws of mathematics say that x/y = z. That's testable, and repeatable. The Pentium in question would say that x/y = z.00000001 or something like that. The flaw was found, and corrected.
So yes, we can make machines, and those machines can be fallible to some degree. But they can also be refined and improved, and their results can be retested by others to get either confirmation or conflict. But at the most basic level, if you're talking about simple logic gates, they will do exactly what they are meant to do: rudimentary logic, at a level of consistency far better than our primate brains can manage.