Are robots our friend or foe?
Are they sinister metal contraptions that will erase mankind or technological wonders that will improve our lives?Josh Sims May 1, 2016
“I’m sorry Dave. I’m afraid I can’t do that."
The words of HAL 9000, the computer in Stanley Kubrick’s 1968 movie 2001: A Space Odyssey, perhaps best sum up the contemporary fear of technology advancing to such a degree that it begins to make decisions for its own, rather than its human master’s benefit. Give it legs, and we’re really in trouble.
Suspicion of the advance of artificial intelligence, especially when embodied in the unnerving form of a humanoid robot, has been the bedrock of big screen sci-fi nightmares. From Blade Runner to Demon Seed to The Terminator, the idea that robots – defined by The Robotics Institute of Carnegie Mellon University in Pittsburgh as “reprogrammable, multifunctional manipulators” – are less home-helpers as home-wreckers is prevalent.
There has always been the fear that domestic robots are just waiting to join metal hands and take over the world. Or, if not take it over, then at least drastically remodel the way it works. Indeed, a 2013 Oxford University study has estimated that machines could eventually replace a huge 47 per cent of the US labour force, while a UK study suggests that 35 per cent of the country’s jobs could be “at risk”.
Take just one industry by way of example: hospitality. Robot drones are now being used to deliver items to guests at the Casa Madrona hotel in California; a robotic luggage storage system has been installed at the Yotel in New York’s Times Square; Botlr, recently introduced at Aloft Hotels, works as a robotic butler, delivering items to guest rooms and accepting tweets in place of tips.
The new Henn-na hotel in Sasebo, Japan, takes things further with its staff of 10 robots and only two humans. Robots can greet guests, carry bags and clean rooms, as well as speaking several languages and responding to guest enquiries.
Depending on one’s viewpoint, robots will herald a dystopian era of global mass unemployment and perhaps new Luddite reprisals, or a utopian one of leisure and drastically reduced working hours.
Yet this is more about automation of any kind diminishing human utility, rather than an active take-over by thinking, moving machines. In fact, Isaac Asimov’s seminal 1950 book I, Robot laid down in fiction the Three Laws of Robotics that have been adopted by many robot developers: that a robot must obey humans, cause no injury to humans and protect itself as long as this is not counter to the other laws. And yet, there is already a history of robots blowing a fuse and killing people. In 1992 a software glitch in a Silicon Valley industrial robot called Robbie (named after Forbidden Planet’s whirring star) led it, Robocop-like, to decapitate its operator.
Of course, like it or not, sophisticated robots are already part of our world, in the form of computerised limbs. They build cars, for instance, are increasingly used in nuclear processing and even assemble computer chips – how’s that for being a ‘self-made man’. In agriculture, a growth area for robots, they can shear sheep and prune grapevines. In medicine they search for brain tumours. They explore the depths of unknown territories, explode bombs and go to Mars.
They are getting smarter too. As of 2012, a Yale University robot was able to look into a mirror and recognise itself, the classic test for self-awareness. Certainly the progress of robotics to the point at which they are computationally superior to humans seems guaranteed.
The human brain may, for the time being, remain hugely more compact and energy efficient than its computer rivals (the brain requires just 12 watts of power to do wondrous things) but, all the same, computers’ speed now allows them not only to beat chess grandmasters, and the more complex Chinese board game Go, but even to win on the US quiz show Jeopardy! which requires a subtle appreciation of more than hard facts.
Moore’s Law, which says that CPU power doubles every 18 to 24 months, and which to date has proven correct, suggests that human equivalents, at least in sheer processing power, will be possible by just 2040. And, according to the acclaimed robotics engineer Hans Moravec, it is this leap in computing that will lead us to overcome a more obvious hurdle to a robotic future: developing robots to such an extent that they are more than mobile, computer-driven machines but can multi-task, interact with the world and cope with the unexpected.
Moravec, who in 1979 created the first autonomous robot able to navigate across a room of obstacles, has predicted – perhaps over-enthusiastically – that universal robots as mentally powerful as small reptiles and programmable for simple chores will be available within a decade, with subsequent decades seeing an evolution not unlike that of biological intelligence, but much faster: at one year in a robot for every 15 million in nature.
And there is no reason to assume that physical evolution won’t be as fast too. Hugo Elias, head of R&D at Shadow Robot, a world-leader in the hugely complex development of human-like robot hands, notes how “the design of the human hand is really just an accident of history, a result of hanging from trees.” It’s not a good design either. “The fourth and little fingers are very weak. The fingers can’t be bent back, which would be useful. The thumb is the only opposable digit against four fingers. Our robot hand has eliminated all of those problems and has more degrees of freedom than a human hand. In some respects its design is definitely an improvement.”
It’s an intriguing, maybe frightening prospect. Yet the robotics community, while certain that the sci-fi vision will eventually materialise, doesn’t expect it any time soon. “Robotics really is an irresistible force,” says Matt Mason of the Robotics Institute, which was founded in part by Moravec. “Computers live in beige boxes and are oblivious to what’s around them. Robotics is about changing that. An animated machine fires the imagination, even if it doesn’t really have awareness yet. But the gold standard in robotics – human performance – may be beyond our ability for a long time. ”
And this is for the basic reason that, issues of brain power aside, building robots is harder than is imagined. “Even if a robot has legs and can stand, what next?” asks Shadow Robot’s managing director Rich Walker. “It needs to be able to do something when it gets wherever it’s going. RND robotics is full of [such] brick wall challenges. There are lots of ideas for domestic robots, for example, which get people excited. When you tell them it’s likely to cost £500,000 a pop, they tend to go ‘Hmm, you can buy a lot of hired help with that’.
"Only 14 per cent of European homes have a dishwasher. By playing ‘what if’ games, science fiction may have given us ideas as to what could be possible, but it has also given the public the idea that building robots is simple.”
Indeed, it is because, in contrast, humans are so ineffably sophisticated – “most of us think that just walking is pretty simple, until you try and get a machine to do it,” says Mason – that philosophers seek to reassure us that robots’ strictures will always keep them subservient to us. A.I. they say, may allow robots to understand how, but will never understand why. The lack of a psyche or soul will be their permanent limitation.
But this is, as Descartes’s ‘man machine’ belief suggests, to assume even humans have that. It is also to discount the idea that robots may be us but without our murderous, ruthless and petty instincts. Robotics may end up giving us profound insights about ourselves. If, that is, we pay attention, because an alternative vision is one of robots so mundane that we will look back on these discussions perplexed by our paranoia.
“The sci-fi view of a world run by machines is very much a sci-fi view,” says Dr. Rob Richardson of the University of Manchester’s Artificial Intelligence Group. “Eventually robots will become what computers became to us in the 1980s. They will just be an everyday thing – nearly everyone will have one.”