A Suicidal Computer? Was Saint Augustine Right?
From Washington, D.C., comes news of the most ingenious proof ever of the existence of God. It comes in the form of an apparent suicide.
National Public Radio reported this weekend on a robot that propelled itself into a fountain, shorting its circuits and effectively ending its life. What if the fate of this robot reflects the real promise, or lack thereof, of artificial intelligence (AI)?
There may be mundane explanations for why the Knightscope K5 security robot ended up in the drink. A programming error may be responsible. Perhaps a routine failure of an electrical circuit did it. Or maybe the robot was somehow tricked or lured into the fountain by a human.
But consider the possibility that the machine’s algorithm made self-destruction seem like the optimal choice. What then?
AI and the search for a replication of the human mind is at the frontier of science and science fiction. If the mind is merely the sum of its logical operations, can’t those operations be broken down, dissected, and then encoded in a supercomputer? Big Blue beat Gary Kasparov in chess; Watson bested the champions of Jeopardy. What can’t some other computer excel at the ordinary tasks of daily living?
The suicidal robot in Washington, nicknamed “Steve” by those familiar with it, was programmed to wander the Washington Harbour complex. Its task? To detect misbehavior by way of thermal image censors and cameras. It had the ability to issue parking citations, for example – no doubt sending electronic notice to offenders. It cost the District of Columbia about $7 per hour to operate, well below the minimum wage.
But for all its smarts, Steve lacked what poets and theologians refer to as a soul. Was it capable of love? Of desire? Could it conceive of ends that make life worth living, or, in robotics, that make energy worth consuming?
The promise of AI is that it will yield machines capable of performing ordinary human tasks in ways that are predictable. But what if a computer could be programmed to learn from experience, as the more sophisticated computers now can. When will computers become super-intelligent, more capable than humankind at solving problems?
Some forward-thinking souls think that super-intelligent machines are only decades away from creation. Will these devices represent an evolutionary advance? What happens when humans are replaced by computer intelligence as the most intelligent things on the planet?
You’ve seen the movies. In the Terminator, the machines decide we are inconvenient, so they arrange a nuclear to rid the planet of the human stain. In the Matrix, humanity is necessary, but only as a lifeform to create the energy necessary to run super-intelligent machines. In these films, super-intelligent machines become capable of conceiving their own ends. They learn to prefer their own survival to those of their creators.
Perhaps that is what the future holds. Plenty of brilliant minds think it so.
I am not so sure.
I reminded all at once of St. Paul’s observation in the Epistle to the Romans: Does the clay say to the potter, “Why did you make me like this?”
Reason, is and always shall be, the slave of the passions, David Hume observed long ago. A computing machine is instrumental reason elevated to sublime levels. But once you’ve programmed a device to solve every conceivable problem, what ends will it seek?
In his playful piece this weekend on National Public Radio, Scott Simon asked “Will … machines begin to wonder: Is this all there is?”
We are what we love, and, near as I can tell, computers don’t love – they reason, calculate, and dance the rhythms their algorithms teach them. But when all that computing power has run its course, what then, what end?
“[M]y mind is clouded by darkness and is far from your face. The road that leads us from you and back to you again is not one that we can measure, or tread with our feet,” Augustine wrote in the Confessions. “[T]here is a part of man that is unknown even to the spirit within him,” he wrote.
AI lacks a soul. How can it commune with other spirits? How can it learn to respond to grace?
While I am certain there is a simpler explanation for Steve’s suicide, I prefer to think of it as a sublime proof of the existence of God. A super-intelligent machine might calculate the metes and bounds of every observable thing and yet miss the one thing alone worth seeing, the small voice within that has the capacity to respond to grace with faith.
Perhaps we have nothing to fear from AI. Perhaps a machine untethered from any sense of the divine is destined to self-destruction. Steve’s death might just be a powerful proof that what makes us human is not what we share with machines, but what machines lack – a soul.
Steve ran reason’s tether to its end. The pot said to the potter, “Why?” Hearing no response, it self-destructed. This seems a proof-text for many a sermon.