Problem:
 a() -> a()
 g(a()) -> b()
 g(a()) -> g(g(a()))
 g(b()) -> a()

Proof:
 Nonconfluence Processor:
  terms: b() *<- g(a()) ->* g(b())
  Qed
  
  first automaton:
   final states: {4}
   transitions:
    b() -> 4*
  
  second automaton:
   final states: {5}
   transitions:
    a() -> 5*
    g(6) -> 5*
    b() -> 6*