Could work (because better decisions were made), or could be terrible (because people stopped feeling responsible).

We should try it :-)

Time to assemble all the evidence in both directions from sci-fi (such as most of Star Trek: TOS, some of Futurama, etc), feed it into a deep learning algorithm and see if it models a better machine society or a worse one. I trust this is a useful first case to test this hypothesis. :)

This algorithm needs to be very crafty to avoid manipulation. We'll have to constantly re-adjust the weights

Math is already in charge of society. More math would be good for all of us