How long before machines are cleverer than us? The first robots were clockwork novelties. Now robots – named for the Russian word robotnic, meaning worker – can outperform humans at ever more tasks.
But robots are specialists. They don’t have our versatility – yet. They can’t yet do what every child can do – transfer learning to a new context. A human trumpeter could make a fist of playing a clarinet, but a robot would have to be reprogrammed. It might be a world-beater at Go but it couldn’t play Snap.
Making robots which can perform as well as humans, whether it be at walking, seeing, or thinking, involves understanding how humans walk, see and think. Nature is hard to emulate. But . . . remember those lumbering movie robots? They’ve upped their game. Now that we understand the complex mechanics of human ambulation it is possible to build robots which walk like we do, even over rough surfaces.
Robots’ ‘eyes’ can follow humans and even mimic their expressions, but they don’t yet see the way we do. When engineers have cracked the saccadic movements of human vision, maybe a housekeeper robot will be able to tell whether that white stuff on the carpet is cocaine or marshmallow and so be able to clean it up without making a worse mess.
Google’s DeepMind has analysed the neural basis of transferring learning and has recently overcome, at least partially, the robots’ problem of ‘catastrophic forgetting’. So future robots will be cleverer. (It is perhaps comforting to know that as they become more human in their capabilities their joints wear out and their backs give them problems.)
Clever robots are now supporting human health professionals, though not yet supplanting them. Robots can perform complex surgical procedures, but only under the control of a human surgeon. They can read mammograms quickly and reliably, but they can’t break bad news to patients. Toyota’s human support robots can pick up a dropped phone and give it to a bedridden client, but can they give a friendly word and ask about the patient’s cat?
Well, they’re getting there. ‘Social robots’ can sometimes succeed where humans fail. Zora is the size of a small person. Institutionalised elderly people take her under their wing and respond better to her encouragement to join activities than the entreaties of a human organiser. Child-sized Zeno has humanoid features. Children with autism find him fascinating and less threatening than coping with human emotions; by interacting with Zeno they can gradually become more comfortable with human social interaction.
But we still don’t really understand consciousness and emotional intelligence – the things that make us human. If we don't understand our own consciousness, how can we build a conscious robot?
Will this gulf be bridged by cybernetic organisms – humans who have integrated technology into their bodies? Despite dictionaries’ definitions, cyborgs are not creatures of science fiction. We have always modified ourselves: tattoos and piercing for adornment or ritual, compensating for deficiencies with spectacles or joint replacements or pacemakers. Neil Harbisson was born with achromatopsia but now ‘hears’ colours via an electronic antenna implanted in his skull. So he’s a cyborg, and as long ago as 2004 he persuaded the UK Passport Office that the device was part of him, so should feature in his passport photograph. How different is this from a cochlea implant? And even a conventional hearing aid can become an extension of its user’s brain – they function as one unit.
Citizen hackers are taking medical developments into their own hands. Several hundred diabetics have found security loopholes in their continuous glucose monitors and insulin pumps. They have inserted their own predictive algorithms to achieve much more physiological control of their diabetes. The industry is now catching up, working on the commercial development of an artificial pancreas.
Modifications to address health problems can be controversial – gene therapy for mitochondrial disease for example, or deep brain stimulation treatment for Parkinson’s or Tourette’s or locked-in syndrome – yet some cyborgs are taking body modification even further. Natural selection generates good-enough solutions to problems, they point out, but why settle for good-enough when we can now improve on nature?
If your colour sense depends not on your retinal cells, but on an electronic device, why not do as Neil Harbisson has and set it to ‘see’ infrared and ultraviolet? Or hack your hearing aid so you can hear Wi-Fi? Or implant devices which give a buzz when you face north, or when there is an earth tremor, or when the wind is in the east?
Why pollute the planet with light if we can improve our night vision?
Our physical and digital identities are already merging, and cyborgs see themselves as the vanguard of the post–human age, when we change ourselves rather than our environment. Why pollute the planet with light if we can improve our night vision? Why risk losing a smart card when you could implant its chip so you pay for your bus journey or your coffee or sign on to the NHS net by waving your hand over the reader? Why stop with electronics? Why not try to improve your intellectual capacity by hacking your gut microbiome?
There will be problems. Humans are so successful because they are versatile. Making yourself exceptional will have trade-offs: blades may enable you to qualify for an Olympic sprint but they aren’t a patch on a multipurpose prosthesis for doing a spot of gardening.
And who’s responsible if citizen hacking goes wrong? If you can tamper with your own device, who else can hack it and take control of your health or your motivation or your affect? Could cyborgs use their transhuman capacities to dominate the rest of us?
We want robots working for us, not us for them. Robots will certainly take on more health care roles, but until they can read suppressed emotions, general practitioners are sure of a job. Meanwhile medical developments and personal experimentation will turn more of us into cyborgs. Look out for surprises.