An allegory Consider a human controlling a very fast car on a busy street using counterfactual oversight. The car is perfectly capable of driving safely. But what happens in the 1% of cases where the car decides to pause and ask the human to review its proposed behavior, or to suggest an action? Without some […]
