I'm not sure that erring on the side of caution is a bad thing when it comes to commercial jets, especially when "erring on the side of caution" means requiring an airline carrying large numbers of paying passengers needing to pay two qualified pilots to sit in a cockpit just in case things go wrong. I suspect most people underestimate just how many different separate electronic systems, usually developed and manufactured by independent specialists, are involved in controlling an aircraft (not to mention the new ones that would need to be developed for when things go wrong: there's a big difference between an autopilot landing an an instrument landing system and a pilot gliding his plane to a perfect stop on the Hudson River.)

Also worth pointing out that the CEO of Google (yes...I know his company tests driverless cars) isn't really more of an expert on the FAA certification process than the average person here. I think he's on the right side of the bet, but I'd also note that the avionics which assist the pilot in controlling the plane and monitoring what's going on have changed more in the last 50 years than any other aspect of commercial aircraft.

Commercial considerations are a bigger factor than certification in new aircraft programme inertia: airlines' economies of scale hugely favour existing models, and expensive radical innovation in aircraft programmes can often be improved upon at lower research cost by rivals.