© Maciej Kulczynski / Global Look Press
Creating robots with anxiety might keep the machines from talking over the world, according to new research into AI technology.
Researchers from the University of California based their findings on a mathematical model that simulated interaction between a robot and a human. With this model, the team found they could lower the robot’s self-confidence, from the unwavering level it has by default, to better obey their human’s orders.
In one simulation, the robot’s ‘mind’ was given the opportunity to turn itself back on after a human had switched it off. While most AI bots would override the command without fail, the robot with installed self-doubt decided to not question the human’s wishes and remain deactivated.
The researchers noted the importance of finding a balance in confidence, as too little self-belief would mean that even children would have the ability to override the robot’s better judgement.
Overall, the Berkeley team believe this new model will better equip bots to integrate into society and live harmoniously amongst us.