Google works to prevent machine insubordination
Google’s DeepMind is hard at work to prevent AIs from taking over the world.
While the world of the Terminator series looks like a grand adventure on screen, real life would make such a future a little hard to bear. To ensure that a machine uprising does not occur, Google is already hard at work in creating the means that would ensure humanity does not fall prey to murderous mechanical constructs.
According to Engadget, Google’s DeepMind has developed a framework that will allow humans continued control over the actions of AI (artificial intelligence) and prevent it from developing countermeasures and revert control to its human handler.
On a practical note, the capacity to “safely interrupt” an AI-driven construct will prove valuable in case of a situation where its initial programming cannot provide it with a programmed response to an unprecedented event.
For example, if a robot were ordered to carry boxes into a warehouse, it would do so under all circumstances. But if it starts raining and the robot is not aware that it might damage its precious cargo, the human controller can “interrupt” the robot and put it on standby until the rain lets up.
It will take a while before DeepMind is able to create a full-proof kill switch, so in the meantime, let’s hope that any robot uprising won’t happen any time soon. Alfred Bayle
