

This, along with the story of the driverless car running over a beloved neighborhood cat, highlight the fundamental problem here: they’re training these systems to be autonomous cars - and they’re impressively close - but they’re still miles away from creating a simulation of an autonomous human. Not everything about operating a car is driving it.








You know I was just thinking that what OpenAI needs to really succeed is someone who knows how to haphazardly throw together a wildly irresponsible, insecure, and unsafe tool that gets really popular really quickly.