>I always wonder why creating synthetic nanorobots to replace organic ones seems more attractive than learning to control what we already have inside.

Would you like to reuse a result of a process that went on for millions of years and reverse engineer all evolutionary accretions created by a random mutation process ? It's not hard to figure out why developers don't like that option, if they work maintaining a large codebase they probably have hands on expirience with that kind of work. It's not something I would unleash on my brain.

My theory is because most sci-fi enthusiasts are not biologists. They simply do not understand how powerful enzymes are and prefer to think about synthetic nanorobots instead. Plus it sounds cooler.

You don't need to reverse engineer your entire body to understand a single enzyme. We do it all the time in biology. Furthermore, even if you create something that goes haywire in an unanticipated scenario, your body has good defense mechanisms against biological abnormalities.

It doesn't know what the hell to do to quantum dots or nano-scale inorganic materials

Good point, however we could learn to make the organic bots do what we need them to without having to learn everything else about how they work - kind of like driving a car without knowing how to disassemble its engine. They're also much more tightly integrated into the existing infrastructure...