Is a "human" machine really smarter?

[Nathan G]

There is an inevitable law throughout the history of computers that continues to frustrate software engineers: the more complex a system is, the more errors are going to creep in, and the worse these errors become. Scientists continue to search (in vain) for a way to combat this problem, but maybe these errors are necessarily unavoidable?

Human minds, which are nearly infinite in complexity, are full of errors all the time. We mumble, we stutter, we complain, we forget; we get confused, or irritated, or tired, or even insane. A computer never experiences any of this, not because it is smarter than us, but because it is simpler than us.

Fast forward to the Singularity. A machine is invented that is so complex, that it develops a human consciousness and self-awareness. However, a result of this complexity is that enormous errors appear that makes any other function useless.

In other words, a computer can either be sentient or intelligent, but never both. AI's that pass the Turing test will only be invented for entertainment purposes, or as an artificial slave class.