(Large scale) robot armies moderately increase my P(doom). And the same for large amounts of robots more generally.
The main mechanism is via making (violent) AI takeover relatively easier. (Though I think there is also a weak positive case for robot armies in that they might make relatively less smart AIs more useful for defense earlier which might mean you don’t need to build AIs which are as powerful to defuse various concerns.)
Usage of AIs in other ways (e.g. targeting) doesn’t have much direct effect particularly if these systems are narrow, but might set problematic precedents. It’s also some evidence of higher doom, but not in a way where intervening on the variable would reduce doom.
(Large scale) robot armies moderately increase my P(doom). And the same for large amounts of robots more generally.
The main mechanism is via making (violent) AI takeover relatively easier. (Though I think there is also a weak positive case for robot armies in that they might make relatively less smart AIs more useful for defense earlier which might mean you don’t need to build AIs which are as powerful to defuse various concerns.)
Usage of AIs in other ways (e.g. targeting) doesn’t have much direct effect particularly if these systems are narrow, but might set problematic precedents. It’s also some evidence of higher doom, but not in a way where intervening on the variable would reduce doom.