None
EN
Extreme concentration of power over ASI has non-obvious advantages
['Seth Herd']
LessWrong
How risky would it be to make powerful AI obey one or a few people?