The Ethics of Robotics in Modern Engineering

A 3,000 word essay on ethical robotics, submitted as part of a university assignment about 'ethics in engineering'. Originally submitted in December 2012.

The Ethics of Robotics in Modern Engineering

 

Ask an engineer to design a high-yield nuclear bomb and he/she may question the ethics of doing so. But ask the same engineer to design a 3D printer �" capable of producing large components with sufficient tolerances for use in machines �" and it may be done without any ethical considerations to the effects on the manufacturing and transport industries which the printer is circumventing. Ethics is an often underrated, yet undeniably integral part of engineering. We must consider the ethical implications of any innovative technology and cast judgment as to whether it is right to proceed. I use the word ‘right’ in this context as a synonym for ‘ethical’, which I use to also (briefly) encompass socioeconomic and political aspects.

One must also realise that there is a fine balance to be found between determining the most ethical courses of action without critically hindering technological progress. In some cases both the pros and cons of a project may have huge implications on society; in these cases it can be very difficult to determine a suitable course of action. A good example of this is nuclear fission; while the deaths of hundreds of thousands of Japanese civilians by the nuclear bombs dropped on Hiroshima and Nagasaki can hardly be justified, the same technology then gave the world nuclear power, which is becoming ever more important as fossil fuel reserves slowly deplete.

Just as the word ‘engineering’ covers a vast range of sciences, fields of study and technologies, so a comprehensive discussion of ethics in all of engineering cannot be condensed into a single, short essay. Thus for the purposes of concision I will be concentrating on the emerging field of robotics. This essay will consider the impact of robots on a wide range of aspects of modern society, particularly focussing on case studies documenting their use in industry, medicine and warfare, before considering a more distant-future perspective and discussing the ethics associated with artificial intelligence.

The work robot was first used in its current form in the English language in 1920 by Czech playwright Karel Čapek as a suggestion from his brother Josef [1]. He used it as an anglicisation of the Czech word robota, meaning ‘serf labour’. The basic concept of a robot has remained largely unchanged since the days of Čapek. Put simply, practical robots are designed to replace humans. Of course this is not a universal truth, as many technologists and scientists use robots as a means of research into advanced computing or as a way of better understanding the abstract notions of intelligence and thought. That said, almost all of the robots in current use have been designed for a specific purpose other than self-research, with a booming market for home robots such as the Roomba autonomous vacuum cleaner. Such has been the exponential growth of the home robot market recently that the global service robot population has been estimated at more than 18 million as of 2011 [2].

While modern robots have a wide range of applications, they mostly perform tasks that fall under the 3 D’s: dull, dirty or dangerous. Heavy industry, for example, qualifies for all three of these categories. Robots on an assembly line (dirty) perform simple, repetitive tasks (dull), often with high-power tools and machinery (dangerous). It is perhaps no surprise, therefore, to see that (high-volume) modern car factories are almost entirely automated; humans’ largest involvement is usually reserved to quality control, where a visual inspection is needed and robot costs begin to exceed human labour costs. Smaller automotive manufacturers, particularly luxury marques, will still hand-build their cars. Smaller production volumes and rates mean automation may be unnecessary, though low production rates and labour costs tend to drive up the costs of these cars.

This ‘revolution’ in manufacturing (if it can be called so) engenders many moral and ethical questions, which can mostly be condensed into one: is it ethical to replace human labourers on a production line with robotic workers? One could justify it from a utilitarian point of view: the costs involved in purchasing and running a robot are considerably lower than those for equivalent human labour output. For example, within a year of buying two industrial robotic arms, Blue Chip Manufacturing in Ohio had already made the money back from the initial purchase in overall savings, and in 4 years had saved an estimated $120,000 in manufacturing costs [3]. Robots are more efficient and less likely to make mistakes �" the possibility of human error becomes zero (assuming no other significant human involvement). The increased reliability and efficiency can also result in higher passenger safety and better build quality in road vehicles.

Does this, however, fully justify automation? While the use of robots will ultimately produce cheaper, safer cars for millions of customers worldwide, this surely is little consolation to the workers who are left jobless. A common counter-argument is to say that such automation frees up the workers to be more productive in other areas where robots are not yet advanced enough to work. In an ideal world this would be the case, but in reality not all workers would be sufficiently qualified or experienced to simply transfer to a different job.

It is important to note how this is simply another iteration of a centuries-old debate dating back to at least the start of the Industrial Revolution in the 1750s. ‘Technological unemployment’, as it has come to be known, is the central argument for preventing technology from reducing the number of human labourers needed, supposedly leading to higher levels of unemployment. Most importantly, while local unemployment may be affected by such technology, on a macroeconomic scale no effect is observed on national unemployment rates. This has led to the use of the term Luddite Fallacy to refer to the false ‘theory’ that technological advances are ultimately injurious to the economy [4]. I conclude this example by saying that in my opinion, the use of industrial robots is entirely justified from an ethical standpoint, as it ultimately benefits the economy and society.

We have so far considered the ethical implications of using robots as human replacements where they follow a set of pre-programmed commands and no more. But we must also consider situations where robots could be relied upon for more. Robots today are increasingly being used in life-or-death situations where their superhuman precision and/or expendability are desirable �" the most obvious examples being surgical procedures and warfare. Looking ahead to the future and the rapidly advancing field of artificial intelligence, the question arises: might we someday give robots the responsibilities of making life-or-death decisions? For example, would we deploy robots in battlefields with the ability to judge whether a person is an ally, enemy, or innocent bystander �" and then decide whether or not to shoot them? Or would we employ a ‘doctor robot’ with the capacity to decide to turn off the life-support system of a terminally-ill patient? Fundamentally, these situations question the trust we place in robots and their ‘intelligence’. In order to best understand the ethics of empowering robots in this way and giving them authority, it is perhaps important to first examine the capabilities of those currently used in medicine and warfare.

In a Boston hospital, a young girl is undergoing an operation to remove a kidney blockage. The inch-long incision has been made in her abdominal wall, and the surgeon is removing the blockage with a scalpel and tweezers �" but he is not standing over her. Instead he is sat a few metres away looking at a screen. He is controlling a da Vinci Surgical System, which is transmitting the live feed to the surgeon’s console via a stereoscopic camera on the end of one of the machine’s arms, while he precisely controls the surgical instruments connected to the other three arms [5]. Performed by hand, this operation would likely leave the girl with a 15cm scar on her abdomen and she would have to spend up to five days recovering in hospital. With the robotic surgeon, she will return home the following day with a regular plaster covering the closed incision. Robotic systems are slowly revolutionising surgical procedures in a way matched only by the 19^th Century introduction of pre-operation instrument sterilisation. Robots like the da Vinci system are significantly increasing the success rates of complex, dangerous and often life-threatening procedures, ranging from delicate cardiac operations to hysterectomies. As these systems are capable of much greater precision than humans, both recovery times and risks of complications are also greatly reduced. But these systems are in no way autonomous. They are called robotic �" and in a sense they are, as they at least partially fulfil the ‘human replacement’ criterion �" but they lack the capacity to function without human input. They are quite literally ‘remote-control surgeons’.

Similarly, none of the estimated 12,000 military ground robots and 48 UCAVs (unmanned combat aerial vehicles) deployed in the Middle Eastern warzones [6] can be described as fully autonomous. Like the da Vinci system they partially replace the human �" they remove the human from the possibility of direct harm or death in a warzone �" but they must always have a human controller inputting commands. Granted, the degree of autonomy in the most advanced military unmanned vehicles is, in some cases, quite high: for example, there exist UAVs which are capable of autonomous flight to the point where a ‘pilot’ need only input a set of map co-ordinates for the UAV to fly to and gather surveillance data. The important point to make, however, is that there is always a human triggerman behind any unmanned vehicle attack. These include ground robots as well as UAVs �" the most basic of which simply comprise a camera and a rifle mounted on a track-driven platform which is remotely-controlled by a soldier. While there are some in development, there are currently no military robots capable of autonomously making a kill decision without human consent.

Consider, though, in the near future, robots that might, in theory at least, be capable of making such decisions. A robot has no emotions. It does not feel empathy, nor does it know compassion. All such a robot would be capable of is applying built-in algorithms to a matrix of pixels, or image, to determine what it should shoot at. Based on this concept, could we ever build a sufficiently complex robot to fully replace humans on the battlefield? Major Daniel Davis (US Army) thinks not: ‘…it is unlikely that any robotic or artificial intelligence could ever replicate the ability of a trained fighting man in combat… Suggesting that within the next 12-plus years technology could exist that would permit life-and-death decisions to be made by algorithms is delusional’ [7]. It is one thing for an artificial intelligence to be able to distinguish between hot and cold bodies or animals and humans (which they have been able to do for some time), but quite another to be able to distinguish between active combatant and civilian non-combatant. The level of complexity associated with such distinctions leads me to agree with Davis. But given the obviously inherent unreliability in predicting future technological advances, we must consider the ethical course of action if such technology were ever created. In a presentation to the advanced weapons department of the US Navy, John Canning addresses the issue of autonomous weapons, and states rather simply: ‘Let men target men. Let machines target other machines’ [8]. This suggestion is an easy escape from all the ethical issues of robots killing humans, or more specifically, the wrong humans.

The case of military robotics is a complicated one ethically, with no single or obvious answer. However I believe that, until the next major paradigm shift in robotics or military technology, Canning’s proposal is probably the best stance to adopt from an ethical point of view.

Perhaps at the opposite end of the spectrum, consider now intelligent robots charged with maintaining our health, instead of killing us. We have already looked briefly at surgical robots, which benefit us by having superior precision and surgical capability in the hands of a trained surgeon. But as the cutting edge of artificial intelligence becomes ever sharper (as it were), are we facing an eventuality of robotic doctors? If feeling unwell, would we seek medical advice from a robot instead of a human? After all, when one goes to a doctor and describes whatever the problem may be, what does the doctor do? He draws upon his knowledge and experience to make an educated estimate of what the ailment may be based on the symptoms, and recommends and appropriate course of treatment. This sort of structured, ordered logic is perfectly programmable into robots (at least in theory, not factoring in present-day technological limitations). It is possible to download Gray’s Anatomy and all medically-related peer-reviewed journal papers into a hard drive; then, when a patient describes his/her symptoms, the robot can search within its database of medical knowledge to diagnose the patient with whatever ailment best fits the symptoms. Of course, as with any artificial intelligence, ‘common sense’ would be lacking - a human doctor would (hopefully) be able to use common sense to determine that a fever-ridden patient is more likely to have common influenza than Ebola, for example. In the future, however, it may also be possible to program a robot with some form of learning capability, in such a way that it could then, through gathered experience, ‘learn’ common sense. I think therefore to have a ‘diagnosis robot’ would be almost ethically acceptable, as long as the patient would not be compelled or forced in any way to follow the course of treatment proposed by the robot.

‘Care Robots’, on the other hand, are a different matter. The concept of these robots is to replace human nurses in hospitals and care homes for ill and/or elderly patients. On the whole, my opinion on these (and other similar ‘service robots’) is this: from an ethical standpoint, they are perfectly acceptable. One could apply the same Technological Unemployment argument as before, but the counter-arguments remain just as valid. The important distinction to make is when these care robots would have life-and-death power over us. Certainly I do not think it is at all ethically acceptable for robots to have the ability to make a conscious decision to ‘pull the plug’ on a terminally ill patient. Such power would be open to misuse and abuse and would raise ‘what if’ questions postulating the possibility of a robot pulling the plug accidentally or even (a sinister thought) deliberately. Though with the increasing fame of the Dignitas euthanasia clinic in Switzerland, another question arises: what if a human asked a robot to end his/her life? I will attempt to answer this in the next section.

Looking ahead into the slightly more distant future, let us now consider the ethical implications of advanced artificial intelligence. While it is currently at a very basic level, it is advancing very quickly. For example, Google recently connected 16,000 processing cores together to create a neural network (or brain), then let it watch Youtube videos for 3 days [9]. After this time, the ‘brain’ had learned how to identify a cat �" among many other things �" all by itself. This is a good example of ‘machine learning’, and is a very important ethical issue to consider. While Google say their computer brain was many orders of magnitude less complex than a human brain, imagine one day when humans create an artificial intelligence to rival our own. Some estimates place that date around the year 2050 [10]. What if, say, these robots then begin to question their subservience to humans and demand to be treated more equally, as beings of equal intelligence? Without entertaining the notion of a Terminator-esque robotic revolution, would we have to give robots rights to go with their responsibilities? Arguably the most famous attempt at devising a set of laws for intelligent robots is the 3 Laws of Robotics, as coined by Isaac Asimov. But would these laws be sufficient �" primarily to protect humans from harm at the hands of robots �" but also vice versa? Asimov uses his many short stories to explore loopholes and shortcomings of these laws, while Prof. Alan Winfield proposes a set of draft ethics for roboticists [11], including such statements as ‘robots should not be designed solely or primarily to kill humans’, and, most importantly, ‘Humans, not robots, are responsible agents’. It is easy to argue, then, that as soon as robots start becoming existential, we could simply build stupider robots, or just stop making those robots altogether. This topic in itself opens up another huge philosophical discussion that begins to stray from the considerations of ethical engineering. That said, it would not be inconceivable to live in a world where intelligent robots are granted similar rights and responsibilities as humans (think of The Bicentennial Man by Isaac Asimov). After all, as biomedical engineering innovations continue to replace faulty organic parts with mechanical prosthetics, the defining line between human and android is becoming blurred.

To conclude, it is obvious that the ethics of robotics will remain a hotly debated topic for many years to come. However I believe that in the majority of cases, replacing humans with robots can be entirely ethically justified. Such ‘upgrades’, if you will, result in increased efficiency and reliability while lowering costs and the risk of human injury or death. Intelligent robots form a separate ethical question, but I hope that as such technological advances happen, legislation and ethical codes will be appropriately updated to prevent any kind of robotic revolution. I personally would quite like to, one day, be able to have an intellectual discussion with a robot, but perhaps this is more wishful thinking than serious prediction.

 

 

References

 

[1]: Čapek, K., 1933. About the Word Robot. Translated from the Czech by N.Comrada. [online] Available from: [Accessed 25 November 2012]

 

[2]: World Robot Population, 2000-2011 [graph] [online] Available from: [Accessed 26 November 2012]

 

[3]: Kimes, M., 2008. Need more workers? Try a robot [online] Available from: [Accessed 27 November 2012]

 

[4]: Toolkit For Thinking: Luddite Fallacy [online] Available from: [Accessed 27 November 2012]

 

[5]: Singer, E., 2010. The Slow Rise of the Robot Surgeon [online]. Available from: [Accessed 28 November 2012]

 

[6]: Gates, R.M., 2011. Remarks by Secretary Gates at the United States Air Force Academy [online]. Available from: [Accessed 28 November 2012]

 

[7]: Davis, D.L., 2007. Who Decides: Man or machine? [online] Available from: [Accessed 28 November 2012]

 

[8]: Canning, J.S., 2006. A Concept of Operations for Armed Autonomous Systems [online] Available from: [Accessed 28 November 2012]

 

[9]: Clark, L., 2012. Google brain simulator identifies cats on Youtube [online] Available from: [Accessed 28 November 2012]

 

[10]: Moravec, H., 2008. Rise of the Robots �" The Future of Artificial Intelligence [online] Available from: [Accessed 28 November 2012]

 

[11]: Winfield, A., 2012. Robotics: A very short introduction. Oxford: Oxford University Press

Author's Note

I shouldn't have to say this, but being British myself this essay is written in 'British English'. Apologies if pointing this out seems insulting, but this is my first submission and I don't want any perceived grammatical errors to distract from the essay's message.

Reviews

[Jocelyn_wonders]

Thank you for writing this essay. I enjoy reading well-written essays and I am curious about the subjects of ethics, robots, and science. I hope you don't mind that I printed this to use as an example of how to cite sources on the internet properly. I've written a number of history and science essays, but haven't posted them because I worry about being sued for plagiarism.
Jocelyn wonders.

[jbtaylor1103]

Thanks for the feedback! I'm thinking of writing a similar essay concentrating on intelligent robots and their rights at some point. And of course, please feel free to use this essay (as long as you reference it if you quote me). I sincerely hope this is the correct reference method, because it's exactly what I was taught at university and my tutor didn't correct me!

[Jocelyn_wonders]

Thank you. I look forward to reading about intelligent robots and their rights!
Jocelyn wonders.