Robots are being applied in fields described as the three Ds, that is, tasks that are dirty, dangerous or dull. A fourth application is use in regions inaccessible to humans such as outer space, remote mountainous or desert land areas, and deep undersea. The jobs most at risk are low skill, routine, highly structured, simple, and rule based, in environments that are static (unchanging) and structured (e.g. on fixed known routes). For example chess is easy for a robot because the operating environment, the chess board, is very simple and the rules are few and easily defined. Whereas ironing and making beds is hard for a robot because of the unlimited combination of shapes, sizes, appearance, and positions of clothes, sheets and blankets. Even walking and avoiding obstacles is hard for present day robots. The relatively high initial cost of robots means that they tend to be used only by large companies or in defence.
It is unlikely that all the humans will be replaced by robots in any task, it is much more likely that fewer people will be needed because for many years robots will be assistants. The future is most likely to be one of productivity gains rather than mass redundancies.
Robots are usually thought of as being mobile but they don’t have to be. Washing machines, dishwashers and tumble driers can also be thought of as robots, particularly the more sophisticated ones that are fitted with microprocessor “brains”, and several sensors and actuators.
The jobs we will see machines taking over first:
1). Train driver. London Underground train drivers are vulnerable as wages are high, strikes are too frequent for travellers, and the skills required to drive the trains are low. The drivers only need to follow a small number of simple instructions. The technology has been available for many years. The Docklands Light Railway is largely unmanned and has been a great success. All that is needed is the political will and the start-up cost for the initial investment. We could see more cybernetic (computer controlled) trains in five years, but ten years is more likely.
2). Taxi driver/goods vehicle driver/chauffeur. In 2004 in the driverless car grand challenge; all the cars failed to complete the course. By early 2014 Google’s self-driving car had travelled 100,000 miles on real roads without a crash. This is an amazing rate of progress. Processing the signals from video cameras and laser rangefinders at speed in very variable and rapidly changing environments and situations is the technical hurdle to overcome. There are health and safety, and legal hurdles to overcome too, and the costs of overcoming them will be significant. The cost of the sensors, the computers and software is high currently. We may see self-driving vehicles on relatively deserted roads in ten to twenty years’ time.
3). Hospital and office porter. Robot “nurses” have been used in the USA for years, partly because the wages of nurses there are sufficiently high to make robots economic. These robots carry patient meals, notes, and mail, distribute bed linen, carry waste, drugs, dressings etc. Because hospital corridors offer structured environments, fetching and carrying in these environments is fairly straightforward to achieve robotically. However the cost of the robot and its programming has to be brought down to the minimum hourly wage for a person to make it economic.
4). Astronaut/space explorer. Robots have already overtaken humans in this field. This is because of the danger, the weight of a person and their life support system, and the need for a human to be returned safely to Earth. It is highly unlikely that humans will ever be used for space exploration beyond the moon.
5). Floor cleaners. Robot floor cleaners have been popular in the home for a few years. Robot lawn mowers have sold in their thousands. It is likely that costs will get lower and performance will increase with continuing evolution and development. Larger scale robotic commercial floor cleaners are in current use in airports and railway terminal concourses. As these machines improve and acquire a good track record they will be used more widely.
6). Teachers and lecturers. The use of computers, television, the internet and computer graphics in education will increase. The Open University has used television and self-teaching texts very successfully. Experiments with teaching machines have also been successful. American universities are teaching courses at BSc and MSc level for students who are not only off campus but not even in the same country. A colleague of mine and I developed is a trio of identical humanoid robots that go into primary schools to teach children the merits of recycling, reusing and reducing waste. The robots (which are partly programmed and partly human controlled) are very effective at teaching the children the value of environmentally sustainable lifestyles. The robots have performed in front of over two million children in the UK.
7). Lab technicians and scientists. Research and development, particularly in biology, often requires tiny, precisely measured volumes of liquid to be added to test tubes, dozens at a time. The enormous task of decoding the human genome was carried out largely by computers and robots.
8). Pharmacists. Robot pharmacists already exist in the UK. The robots pick the right packet of drugs from a huge number of pigeon holes and pass them via a conveyer belt to a pharmacist for checking before handing on to the patient. Thus a busy pharmacy can be run by one pharmacist and a robot.
9). Security guards. Patrolling warehouses, factories and offices etc. can be dull and dangerous. Such environments are usually structured, for example there may be set routes along corridors. They are also generally static, that is, without people moving around in front of the robot. Such environments are relatively simple for robots to navigate.
10). Airline pilots. Passenger aircraft are now so safe: structurally, mechanically, electronically and in computer control that we are approaching the point where most aircraft accidents are due to human error. We are at the point where aircraft should be robots or rather computer controlled, that is they could be allowed to take off, fly, and land by themselves. However most passengers would not accept this and would still like the idea of humans in overall control whatever the accident statistics imply. Airline pilots are likely to be increasingly side-lined and just taking over manual control in situations for which the computer has not been programmed. In perhaps ten to twenty years the pilot will probably still be on board but the computer will be in overall control and able to override the pilot rather than vice versa.
The impact of robots on policing and defence
Policing is such an infinitely varied task that robots will not be up to the task for twenty to fifty years or more. However certain specific tasks such as approaching a terrorist who might be wearing a suicide vest is so hazardous that it would be a task better suited to a robot. Similarly searching and clearing a building with one or more gunmen inside, for whom suicide may be a goal is extremely hazardous for policemen and soldiers. Robots have been developed for these tasks and the evolution will continue. Many modern military aircraft are designed to be highly manoeuvrable while being almost invisible to radar. The resulting design compromises make the aircraft too unstable to fly by any human pilot. The planes are flown by computers, with the pilot instructing the computers. The technology has advanced to such a stage that pilotless aircraft, unmanned air vehicles UAVs or drones are now being used by over eighty countries. It has been reported that there are now more UAVs in the US Air Force than piloted planes.
Robots and medical surgery
There are a few surgical operations that require so many precisely made incisions that they are better suited to robots. In cancer cases it is vital to remove all the cancer cells, but removal of adjacent healthy cells can impair the patient significantly. In some brain operations cancerous cells are located by an MRI scan and the locations are transmitted directly to the robot surgeon. Prostate surgery can also be done more successfully by robot in some cases because of the greater precision of the machine.
Robots in the home
General purpose domestic robots are highly unlikely to appear in the next fifty years. This is because those tasks that we might want them for, ironing, making beds, climbing stairs etc are all hard problems that would need expensive solutions. A robot that could do just one of these tasks would cost as much as a luxury car. Robots that could do one or two very simple tasks such as monitor the health or provide company for an elderly or infirm are however very likely in the next ten to twenty years. Some basic companion robots already exist.
At the moment, nearly every household could have one or more single purpose robots such as floor cleaners and lawnmowers. However I cannot envisage a time in the next fifty years when nearly every household will have a general purpose robot because the cost of such a machine would be so much more than a human maid, au pair, housekeeper or carer.
Robots and human emotions
In the future, robots will have emotional intelligence. Or to be more precise, robots will be able to show most, if not all, of the signs and behaviours of emotional intelligence. The robots will not feel, but like actors they will be able to show emotional intelligence. I would expect robots to develop levels of emotional intelligence that will be greater than many humans. The various aspects of emotional intelligence I would expect robots to be better at than many humans include: thinking, judgement, problem solving, openness, self-control, self-analysis, and paying attention to and identifying people’s feelings and emotions, for example.
But robots wouldn’t be accepted by most people as a replacement for a partner or child. Though some people do accept surrogate partners, for example dolls and pets, for the vast majority of people a robot would fall so far short of a human that this would not be a possibility, even perhaps in a hundred years or more. The cost of producing such a machine would be more than many companies could afford, let alone individuals. Such technology is probably more than fifty years away. But with advances in artificial hearts, lungs, eyes, ears, skin and so on. The creation of a part machine part human is possible technically. Such machines would be cyborgs or cybernetic organisms. Patients have been given artificial brain, heart, cochlea and other implants in large numbers.
The danger of robots to the human race
Everything that is useful has within it the potential to be abused. So there are dangers inherent in robots; just as there are with motor cars, trains, ships, aeroplanes etc. On balance all these technologies have proved to be good things. It will be so with robots. It has been said that when robots become more intelligent than us, if we are lucky, they might keep us as pets. If we are unlucky they might treat us in the same way as we treat chickens, pigs and cows. There is a potential risk that corrupt dictators will create armies of killer robots. Provided the democracies stay technically ahead, as with other potential military technology, there will be an overall benefit to humanity from robots. The bigger risk will be if the democracies allow the dictatorships to become technologically more advanced. Problems or ill effects can be curtailed as they arise on a case by case, trial and error basis. We must be careful not to stifle advances as legislators did with the advent of the motor car when people were required to walk in front of the cars waving red flags.
New TV series Almost Human airs on Tuesdays at 9pm on Watch, premiering on 6th May.
By Martin Smith, Professor of Robotics, Middlesex University