I’m not a programmer by any means, but it seems to me that it might indeed be useful if an AI can generalize from following directions where you want to go right at an intersection to reach your destination, so [opposite] that to left on the way back, to situations where you accelerated to a hundred meters per second to reach your traveling speed, so [opposite] that to decelerate to a stop so you don’t crash at your destination. Or if you heated a sample up by 50 degrees celsius, [opposite] that by cooling it 50 degrees celsius to reach the original temperature. Or, if it can quantify a person’s happiness, then if they lost 50 hedons when their puppy was run over, then [opposite] that to bring them back to their previous satisfaction level.
I’m not a programmer by any means, but it seems to me that it might indeed be useful if an AI can generalize from following directions where you want to go right at an intersection to reach your destination, so [opposite] that to left on the way back, to situations where you accelerated to a hundred meters per second to reach your traveling speed, so [opposite] that to decelerate to a stop so you don’t crash at your destination. Or if you heated a sample up by 50 degrees celsius, [opposite] that by cooling it 50 degrees celsius to reach the original temperature. Or, if it can quantify a person’s happiness, then if they lost 50 hedons when their puppy was run over, then [opposite] that to bring them back to their previous satisfaction level.