"I think the figure for pilot error causing accidents in GA is 83% (NTSB figures, don't know about the UK)."
Do you have the reference, SS?
I don't doubt you, but I'd like to see the breakdown of the other 17%.
The only real worry in flying is loss of control surface(s) or structural failure. The other stuff, e.g. engine failure, is adequately covered by having a second chance, and the probability of two unconnected systems failing concurrently is unbelievably low. An engine failure over land means a forced landing; I think about 80% of those are without injury. For example flying over water is covered by carrying a raft (the chance of an engine failure AND an in-service-interval raft failing to open is near zero). Even flying over mountains above an overcast layer is covered by carrying a GPS with terrain data and being high enough - much as that will make some people cringe.
17% is an awful lot for structural failure in an aircraft operated within design limits. If someone flies into a CB at Vne+20kt and it breaks up, I'd call that "pilot error" too.