> RL researchers have continued to study the possibilities around imagination replay
I think of what is for me the “holy grail” of AI research and development: counter factual reasoning. I think Gary Marcus has it right: “Computer systems need to understand time, space and causality. Right now they don’t.“ To me, causality and counter factual reasoning are similar in that when we get one working we will get the other.
> RL researchers have continued to study the possibilities around imagination replay
I think of what is for me the “holy grail” of AI research and development: counter factual reasoning. I think Gary Marcus has it right: “Computer systems need to understand time, space and causality. Right now they don’t.“ To me, causality and counter factual reasoning are similar in that when we get one working we will get the other.