Will Efficiency-Minded Robots Determine Humans are Obsolete?
March 4, 2015
Sony’s upcoming film ‘CHAPPiE’ presents a world where self-sufficient robots parade around as our reluctant keepers. In a tyrannical police state where a group of rebels’ only hope is to repurpose a stolen instrument of their oppression to fight back, much of the dystopian future illustrated on screen is grounded in the reality we face today. In an earlier piece, I discussed the complicated relationships that arise when human morality is faced with seemingly sentient artificial intelligence. In this post, I would like to discuss the potential threats that may arise through the seemingly innocuous progression of artificial intelligence.
In his book, Superintelligence: Paths, Dangers, Strategies, author and director of the Future of Humanity Institute at the University of Oxford, Nick Bostrom suggests that in the case for A.I., “when dumb, smarter is safer; yet when smart, smarter is more dangerous. There is a kind of pivot point, at which a strategy that has previously worked excellently suddenly starts to backfire.” Bostrom calls this phenomenon, “the treacherous turn”. He argues, while weak, an A.I. behaves cooperatively, even increasingly so, as it gets smarter. However, when the A.I. becomes sufficiently strong, without warning or provocation, it strikes, and begins to “optimize” the world according to the criteria implied by its values.
Such ‘treacherous turns’ may not be as obvious as biding-time to garnering enough strength to overthrown their human masters. A sneaky A.I. may purposely malfunction in order to have its human programmers develop new and somewhat augmented A.I. architecture that it has calculated will be better served to achieve its intended goal. In this respect, A.I. may be seemingly indifferent to its own demise, so long as its goal comes to fruition. Human engineers may appear to gleam further insight into the inner-workings of the machine and install further trust in to the robot, only to have their efforts used against them (so if you were hoping Sony’s termination of their line of Aibo-robots was to thwart a A.I. takeover, they may be playing us like a fiddle!)
In the film ‘CHAPPiE’, humans capture and reprogram a law-enforcement robot to use as a tool to fight for their freedom. The robot (CHAPPiE) appears to be sentient, and is permitted to learn and grow, similar to a developing human-being, rather than in accordance to the pre-destined rules for which it was designed. The narrative is counter to Bostrom’s hypothesis, that when we ‘let the A.I. out of its sandbox’ we march boldly ‘toward our doom.’ In this case, when we let go of the reigns and let A.I. “think” for itself, it turns toward benevolence rather than malice. I certainly hope this turns to an instance of when art imitates life.