I see your point, but I would suggest that human history is full of opportunities to have bemoaned that "There is no evidence we can build X" shortly before someone went and built the first X.
Even absent a general AI in the near future, the rapid increases in automation from ML and weak AI that we're already seeing don't look to reduce in pace or sophistication.
Even absent a general AI in the near future, the rapid increases in automation from ML and weak AI that we're already seeing don't look to reduce in pace or sophistication.