Wednesday evening I sat down to watch the second part of the BBC 4 documentary, Hyperevolution: the Rise of the Robots, in which the evolutionary biologist Ben Garrod and the electronics engineer Prof. Danielle George trace the development of robots from the beginning of the 20th century to today. I blogged about the first part of the show on Tuesday in a post about another forthcoming programme on the negative consequences of IT and automation, Secrets of Silicon Valley. The tone of Hyperevolution is optimistic and enthusiastic, with one or two qualms from Garrod, who fears that robots may pose a threat to humanity. The programme states that robots are an evolving species, and that we are well on the way to developing true Artificial Intelligence.
Last week, Garrod went off to meet a Japanese robotics engineer, whose creation had been sent up to keep a Japanese astronaut company of the International Space Station. Rocket launches are notoriously expensive, and space is a very, very expensive premium. So it was no surprise that the robot was only about four inches tall. It’s been designed as a device to keep people company, which the programme explained was a growing problem in Japan. Japan has a falling birthrate and thus an aging population. The robot is programmed to ask and respond to questions, and to look at the person, who’s speaking to it. It doesn’t really understand what is being said, but simply gives an answer according to its programming. Nevertheless, it gives the impression of being able to follow and respond intelligently to conversation. It also has the very ‘cute’ look that characterizes much Japanese technology, and which I think comes from the conventions of Manga art. Garrod noted how it was like baby animals in having a large head and eyes, which made the parents love them.
It’s extremely clever, but it struck me as being a development of the Tamagotchi, the robotic ‘pet’ which was all over the place a few years ago. As for companionship, I could help thinking of a line from Andrei Tarkovsky’s epic Solaris, based on the novel by the Polish SF writer, Stanislaw Lem. The film follow the cosmonaut, Kris, on his mission to a space station orbiting the planet Solaris. The planet’s vast ocean is alive, and has attempted to establish contact with the station’s crew by dredging their memories, and sending them replicas of people they know. The planet does this to Kris, creating a replica of a former girlfriend. At one point, pondering the human condition in a vast, incomprehensible cosmos, Kris states ‘There are only four billion of us…a mere handful. We don’t need spaceships, aliens…What man needs is man.’ Or words to that effect. I forget the exact quote. I dare say robots will have their uses caring for and providing mental stimulation for the elderly, but this can’t replace real, human contact.
George went to America to NASA, where the space agency is building Valkyrie to help with the future exploration of Mars in 2030. Valkyrie is certainly not small and cute. She’s six foot, and built very much like the police machines in Andrew Blomkamp’s Chappie. George stated that they were trying to teach the robot how to walk through a door using trial and error. But each time the machine stumbled. The computer scientists then went through the robot’s programming trying to find and correct the error. After they thought they had solved it, they tried again. And again the machine stumbled.
George, however, remained optimistic. She told ‘those of you, who think this experiment is a failure’, that this was precisely what the learning process entailed, as the machine was meant to learn from its mistakes, just like her own toddler now learning to walk. She’s right, and I don’t doubt that the robot will eventually learn to walk upright, like the humanoid robots devised by their competitors over at DARPA. However, there’s no guarantee that this will be the case. People do learn from their mistakes, but if mistakes keep being made and can’t be correctly, then it’s fair to say that a person has failed to learn from them. And if a robot fails to learn from its mistakes, then it would also be fair to say that the experiment has failed.
Holy Joe Smith! I was also a reminded of another piece of classic SF in this segment. Not film, but 2000 AD’s ‘Robohunter’ strip. In its debut story, the aged robohunter, Sam Slade – ‘that’s S-L-A-Y-E-D to you’ – his robometer, Kewtie and pilot, Kidd, are sent to Verdus to investigate what has happened to the human colonists. Verdus is so far away, that robots have been despatched to prepare it for human colonization, and a special hyperdrive has to be used to get Slade there. This rejuvenates him from an old man in his seventies to an energetic guy in his thirties. Kidd, his foul mouthed, obnoxious pilot, who is in his 30s, is transformed into a foul-mouthed, obnoxious, gun-toting baby.
The robot pioneers have indeed prepared Verdus for human habitation. They’ve built vast, sophisticated cities, with shops and apartments just waiting to be occupied, along with a plethora of entertainment channels, all of whose hosts and performers are robotic. However, their evolution has outpaced that of humanity, so that they are now superior, both physically and mentally. They continue to expect humans to be the superiors, and so when humans have come to Verdus, they’ve imprisoned, killed and experimented on them as ‘Sims’ – simulated humans, not realizing that these are the very beings they were created to serve. In which case, Martian colonists should beware. And carry a good blaster, just in case.
Garrod and George then went to another lab, where the robot unnerved Garrod by looking at him, and following him around with its eye. George really couldn’t understand why this should upset him. Talking about it afterwards, Garrod said that he was worried about the threat robots pose to humanity. George replied by stating her belief that they also promise to bring immense benefits, and that this was worth any possible danger. And that was the end of that conversation before they went on to the next adventure.
George’s reply isn’t entirely convincing. This is what opponents of nuclear power were told back in the ’50s and ’60s, however. Through nuclear energy we were going to have ships and planes that could span the globe in a couple of minutes, and electricity was going to be so plentiful and cheap that it would barely be metered. This failed, because the scientists and politicians advocating nuclear energy hadn’t really worked out what would need to be done to isolate and protect against the toxic waste products. Hence nearly six decades later, nuclear power and the real health and environmental problems it poses are still very much controversial issues. And there’s also that quote from Bertrand Russell. Russell was a very staunch member of CND. When he was asked why he opposed nuclear weapons, he stated that it was because they threatened to destroy humanity. ‘And some of us think that would be a very great pity’.
Back in America, George went to a bar to meet Alpha, a robot created by a British inventor/showman in 1932. Alpha was claimed to be an autonomous robot, answering questions by choosing appropriate answers from recordings on wax cylinders. George noted that this was extremely advanced for the time, if true. Finding the machine resting in a display case, filled with other bizarre items like bongo drums, she took an access plate off the machine to examine its innards. She was disappointed. Although there were wires to work the machine’s limbs, there were no wax cylinders or any other similar devices. She concluded that the robot was probably worked by a human operator hiding behind a curtain.
Then it was off to Japan again, to see another robot, which, like Valkyrie, was learning for itself. This was to be a robot shop assistant. In order to teach it to be shop assistant, its creators had built an entire replica camera shop, and employed real shop workers to play out their roles, surrounded by various cameras recording the proceedings. So Garrod also entered the scenario, where he pretended to be interested in buying a camera, asking questions about shutter speeds and such like. The robot duly answered his questions, and moved about the shop showing him various cameras at different prices. Like the robotic companion, the machine didn’t really know or understand what it was saying or doing. It was just following the motions it had learned from its human counterparts.
I was left wondering how realistic the role-playing had actually been. The way it was presented on camera, everything was very polite and straightforward, with the customer politely asking the price, thanking the assistant and moving on to ask to see the next of their wares. I wondered if they had ever played at being a difficult customer in front of it. Someone who came in and, when asked what they were looking for, sucked their teeth and said, ‘I dunno really,’ or who got angry at the prices being asked, or otherwise got irate at not being able to find something suitable.
Through the programme, Japanese society is held up as being admirably progressive and accepting of robots. Earlier in that edition, Garrod finished a piece on one Japanese robot by asking why it was that a car manufacturer was turning to robotics. The answer’s simple. The market for Japanese cars and motorcycles is more or less glutted, and they’re facing competition from other countries, like Indonesia and Tokyo. So the manufacturers are turning to electronics.
The positive attitude the Japanese have to computers and robots is also questionable. The Japanese are very interested in developing these machines, but actually don’t like using them themselves. The number of robots in Japan can easily be exaggerated, as they include any machine tool as a robot. And while many British shops and businesses will use a computer, the Japanese prefer to do things the old way by hand. For example, if you go to a post office in Japan, the assistant, rather than look something up on computer, will pull out a ledger. Way back in the 1990s someone worked out that if the Japanese were to mechanise their industry to the same extent as the West, they’d throw half their population out of work.
As for using robots, there’s a racist and sexist dimension to this. The Japanese birthrate it falling, and so there is real fear of a labour shortage. Robots are being developed to fill it. But Japanese society is also extremely nationalistic and xenophobic. Only people, whose parents are both Japanese, are properly Japanese citizens with full civil rights. There are third-generation Koreans, constituting an underclass, who, despite having lived there for three generations, are still a discriminated against underclass. The Japanese are developing robots, so they don’t have to import foreign workers, and so face the problems and strains of a multicultural society.
Japanese society also has some very conservative attitudes towards women. So much so, in fact, that the chapter on the subject in a book I read two decades ago on Japan, written by a Times journalist, was entitled ‘A Woman’s Place Is In the Wrong’. Married women are expected to stay at home to raise the kids, and the removal of a large number of women from the workplace was one cause of the low unemployment rate in Japan. There’s clearly a conflict between opening up the workplace to allow more married women to have a career, and employing more robots.
Garrod also went off to Bristol University, where he met the ‘turtles’ created by the neuroscientist, Grey Walter. Walter was interested in using robots to explore how the brain functioned. The turtles were simple robots, consisting of a light-detecting diode. The machine was constructed to follow and move towards light sources. As Garrod himself pointed out, this was like the very primitive organisms he’d studied, which also only had a light-sensitive spot.
However, the view that the human brain is really a form of computer have also been discredited by recent research. Hubert L. Dreyfus in his book, What Computers Still Can’t Do: A Critique of Artificial Intelligence, describes how, after the failure of Good Old Fashioned A.I. (GOFAI), computer engineers then hoped to create it through exploring the connections between different computing elements, modelled on the way individual brain cells are connected to each by a complex web of neurons. Way back in 1966, Walter Rosenblith of MIT, one of the pioneers in the use of computers in neuropsychology, wrote
We no longer hold the earlier widespread belief that the so-called all-or-none law from nerve impulses makes it legitimate to think of relays as adequate models for neurons. In addition, we have become increasingly impressed with the interactions that take place among neurons: in some instances a sequence of nerve impulses may reflect the activities of literally thousands of neurons in a finely graded manner. In a system whose numerous elements interact so strongly with each other, the functioning of the system is not necessarily best understood by proceeding on a neuron-by-neuron basis as if each had an independent personality…Detailed comparisons of the organization of computer systems and brains would prove equally frustrating and inconclusive. (Dreyfus, What Computers Still Can’t Do, p. 162).
Put simply, brain’s don’t work like computers. This was written fifty years ago, but it’s fair to ask if the problem still exists today, despite some of the highly optimistic statements to the contrary.
Almost inevitably, driverless cars made their appearance. The Germans have been developing them, and Garrod went for a spin in one, surrounded by two or three engineers. He laughed with delight when the car told him he could take his hands off the wheel and let the vehicle continue on its own. However, the car only works in the comparatively simply environment of the autobahn. When it came off the junction, back into the normal road system, the machine told him to start driving himself. So, not quite the victory for A.I. it at first appears.
Garrod did raise the question of the legal issues. Who would be responsible if the car crashed while working automatically – the car, or the driver? The engineers told him it would be the car. Garrod nevertheless concluded that segment by noting that there were still knotty legal issues around it. But I don’t know anyone who wants one, or necessarily would trust one to operate on its own. A recent Counterpunch article I blogged about stated that driverless cars are largely being pushed by a car industry, trying to expand a market that is already saturated, and the insurance companies. The latter see it as a golden opportunity to charge people, who don’t want one, higher premiums on the grounds that driverless cars are safer.
Garrod also went to meet researchers in A.I. at Plymouth University, who were also developing a robot which as part of their research into the future creation of genuine consciousness in machines. Talking to one of the scientists afterwards, Garrod heard that there could indeed be a disruptive aspect to this research. Human society was based on conscious decision making. But if the creation of consciousness was comparatively easy, so that it could be done in an afternoon, it could have a ‘disruptive’ effect. It may indeed be the case that machines will one day arise which will be conscious, sentient entities, but this does not mean that the development of consciousness is easy. You think of the vast ages of geologic time it took evolution to go from simple, single-celled organisms to complex creatures like worms, fish, insects and so on, right up to the emergence of Homo Sapiens Sapiens within the last 200,000 years.
Nevertheless, the programme ended with Garrod and George talking the matter over on the banks of the Thames in London. George concluded that the rise of robots would bring immense benefits and the development of A.I. was ‘inevitable’.
This is very optimistic, to the point where I think you could be justified by calling it hype. I’ve said in a previous article how Dreyfus’ book describes how robotics scientists and engineers have made endless predictions since Norbert Wiener and Alan Turing, predicting the rise of Artificial Intelligence, and each time they’ve been wrong. He’s also described the sheer rage with which many of those same researchers respond to criticism and doubt. In one passage he discusses a secret meeting of scientists at MIT to discuss A.I., in which a previous edition of his book came up. The scientists present howled at it with derision and abuse. He comments that why scientists should persist in responding so hostilely to criticism, and to persist in their optimistic belief that they will eventually solve the problem of A.I., is a question for psychology and the sociology of knowledge.
But there are also very strong issues about human rights, which would have to be confronted if genuine A.I. was ever created. Back in the 1970s or early ’80s, the British SF magazine, New Voyager, reviewed Roderick Random. Subtitled, ‘The Education of a Young Machine’, this is all about the creation of a robot child. The reviewer stated that the development of truly sentient machines would constitute the return of slavery. A similar point was made in Star Trek: The Next Generation, in an episode where another ship’s captain wished to take Data apart, so that he could be properly investigated and more like him built. Data refused, and so the captain sued to gain custody of him, arguing that he wasn’t really sentient, and so should be legally considered property. And in William Gibson’s Neuromancer, the book that launched the Cyberpunk SF genre, the hero, Case, finds out that the vast computer for which he’s working, Wintermute, has Swiss citizenship, but its programming are the property of the company that built it. This, he considers, is like humans having their thoughts and memories made the property of a corporation.
Back to 2000 AD, the Robusters strip portrayed exactly what such slavery would mean for genuinely intelligent machines. Hammerstein, an old war droid, and his crude sidekick, the sewer droid Rojaws and their fellows live with the constant threat of outliving their usefulness, and taking a trip down to be torn apart by the thick and sadistic Mek-Quake. Such a situation should, if it ever became a reality, be utterly intolerable to anyone who believes in the dignity of sentient beings.
I think we’re a long way off that point just yet. And despite Prof. George’s statements to the contrary, I’m not sure we will ever get there. Hyperevolution is a fascinating programme, but like many of the depictions of cutting edge research, it’s probably wise to take some of its optimistic pronouncements with a pinch of salt.
Beastrabban\'s Weblog 