Understanding shared_ptr vis a vis bst

I was trained in "C with Classes" style C++ and I need to rapidly get acclimated to modern C++. I am writing a very simple bst with shared_ptr

 class bst
 {
 private:
     struct node
     {
         int data;
         shared_ptr<node> l, r;
     };
     shared_ptr<node> head;
     int size;


 public:
     bst() { size = 0; }
     shared_ptr<node> doinsert(shared_ptr<node> curr, int n)
     {
         if (curr.get() == nullptr)
         {
             curr.reset(new node);
             curr->data=n;
             size++;
         }
         else if (curr->data > n)
             curr->l.reset(doinsert(curr->l,n).get());
         else if (curr->data < n)
             curr->r.reset(doinsert(curr->r,n).get());
         return curr;
     }

     void insert(int n)
     {
         shared_ptr<node> temp = doinsert(head,n);
         head.reset(temp.get());
     }

     inline int getsize() { return size; }
 };

And I have a simple harness:

 int main()
 {
      vector<int> v = {2,1,1,2,3,1};
      bst temp;
      for(auto i : v) temp.insert(i);
 }

It looks like the first element gets inserted just fine but things go fubar after that. It wouldn't be resetting the pointer would it?