Superintelligence

Roman Yampolskiy sounds radical because he reverses the burden of proof: critics do not have to prove that superintelligence will become dangerous. The labs have to prove that they can control it.