Oh I agree. There is such a state in computers now but it’s considered “Racing Conditions”.
Remember: Computers are modeled after a human’s understanding of how the brain works. The brain is not modeled after a computer.
With proper usage of race conditions (so that they cease to exist and are not preprogrammed, and _not_ predictable) – and enough self-referential systems, I don’t see any reason why a computer couldn’t be self-aware at some level.
Still, I’m not worried about AI. If it happens at all, it’s hundreds of years from now. We always over-estimate our abilities as humans and think we know more than we do.