Robot Driving at Dusk

A World Where Robots Are In Control

Group 68

Autonomous vehicles (AV’s), once the thing of fantasies has become a thing of reality – virtually overnight. In an age of rapid advances in technology it is easy to lose sight of the essential questions; is the development of self-driving cars ethically justified?

This blog will dissect and analyze the different ethical viewpoints for self-driving cars in a ‘For and Against’ format, therefore allowing the reader to come to an informed decision on the matter.

[AI]ccelerating the Future

Robot Driving at DuskGreater efficiency, safety, as well as removing the headache of driving are just a few of the reasons why development in the AV industry is flourishing. According to a study done by Harvard Health Watch, the average person spends 296 hours driving annually, that’s more than 12 DAYS a year spent behind the wheel!

The ethical considerations of AI driven cars are well contested but from a hedonistic point of view, the benefit of AV’s can be considered more than justified. Studies have shown that driving in general, whether it be a daily commute to work or a long journey to meet family, is linked with higher obesity rates , depression and stress. Hedonism argues that it is the right of the individual to be able to do everything in their power to achieve maximum happiness or pleasure – net pleasure being defined as pleasure minus pain.

People consider the daily drive to work more of a “pain” than a pleasure and the implementation of AI cars would do much to reverse this.

Waymo (Google’s Self Driving Car Program) has been tested for over 2 million miles of autonomous driving in the U.S. and out of 18 accidents, only one accident was the fault of the car – a result of poor road conditions. The statistics show that Waymo has an accident rate of 10 times lower than the safest demographic of human drivers (60-69 years old) and 40 times lower than new drivers. Driverless cars are very thoroughly tested, and the statistics speak for themselves.

An ethical argument commonly made against AVs is that in the future they may be making difficult moral decisions in choosing the lesser of two evils, but one must ask themselves, is leaving the outcome to fate really a morally better option?

A series of surveys made by Prof. Iyad Rahwan from MIT were conducted in 2015 in which stakeholders were presented with different ‘sacrifice scenarios’, they had to determine the best course of action. The common moral attitude from the experiment was that the AV’s should swerve, opposing the view that it should be left to fate. The results showed that most people (76%) adopted the utilitarian approach to safety ethics.

Utilitarianism states that the action which maximises utility or minimises total harm is the correct one. Jeremy Bentham, the founder of Utilitarianism explained it as the sum of all the happiness which results from an action, minus all the suffering involved. This study showed that though there may be some difficult questions which need answering, there is an agreed structured approach which can be taken to answer some of the most difficult questions.

License to Kill

It is undeniable that there will be situations where self-driving cars will face a situation where they must decide on who to kill. Those who have an eager interest in following the development of self-driving cars will be aware of the classic trolley thought experiment, a trolley is hurtling down a track towards 5 people but there is an opportunity to switch the track to a path with only one person, what would you do?

The Trolley ProblemIt can be argued from an Act Consequentialist viewpoint that the consequences of each outcome should be looked at and the good maximized. In the case of Act Consequentialism, the ‘good’ is defined as human welfare and so in simple terms the consequence which maximises human welfare should be chosen i.e. the track should be switched. However, this raises a myriad of issues, would you choose to kill 1 baby over 5 old people?

The complexity and varying nature of this issue makes subjectively ‘incorrect’ decision making inevitable and could incur heavy costs on stakeholders – both car owners and car companies. Less individuals would want to buy a car which may at some point make a very costly mistake. On December 7th , 2017 a motorcyclist was knocked to the ground by an autonomous vehicle and the first lawsuit of its kind ensued. There were no fatalities in this case but imagine the consequences when systems are actively participating in choosing who to kill.

As was shown in the ‘Fast and Furious 8’ movie, self-driving cars would be very much vulnerable to cyber-attacks by hackers, though only a film concept, there are real safety concerns around this issue. Director of KPMG’s cyber security team, Wil Rockall warns that spam jams and hacker-driven congestion may influence self-driving experience in the future. Network data security is very important for ordinary citizens, especially government officials. Protection of personal information and security are very real concerns which need answering, who’s responsibility will it be to counter cyber threats to AV’s? – no doubt it will be a costly undertaking!

Social contract theory states a type of rule-ethical egoism stating that all our actions are selfishly motivated, people are better off living in a world with moral rules than one without moral rules. It outlines how we are subject to the whims of other people’s selfish interests in the absence of moral rules, in this case, the hackers. By adopting AV’s, we would naively be working towards a road system in which implementing moral rules would be much more difficult hence creating an opportunity for those with more selfish interests to take advantage. Would you feel safe in the seat of an AV car which could lose control at any moment?

59 thoughts on “A World Where Robots Are In Control

  1. Could you set the switch to halfway and could that derail the trolley? 🙂

    This is a good article that contains a good level of ethical consideration. It’s a very interesting topic, AVs are (so far as is proven) safer, but are AVs something that the majority of human beings want? Although many people’s perception of AVs is wrong, the issue is will AVs make the majority of people happy? AVs may make us safer but the restriction in choice may reduce happiness.

    A difficult question to answer, therefore a good ethical topic.

  2. From my perspective view, the autonomous car would be increasingly popular in future, but the safety must be the most controversial problem. Although there exist some possibility to happen accident, it is safer compared with traditional Manual driving, because it has several different sensors and supervising all directions of surroundings. Meanwhile, there is only one accident during thousands of tests. Technically and statistically, the autonomous car is safer. If the autonomous car kills somebody, it always makes people feel like that technology kill somebody instead of an accident. It’s hard to make people change this idea.
    Elon Musk said, “If you’re sceptical of self-driving technology, you’re killing people”. Do you agree?

    1. I think you have a very good point, it should be understood that self-driving cars will save more lives than killing them. The ethics behind decision-making shouldn’t detract from the fact that AV’s are safer than manual drivers.

  3. Autopilot is getting more and more attention, and more and more people are enjoying the convenience it brings. However, this system is not very mature and cannot fully replace the driver’s operation. Therefore, people should be cautious.

      1. I agree that it could be boring but that’s because you’re probably sitting there doing nothing making sure the car doesn’t make a mistake. Imagine if you could trust the car then you could get on with some important work or even do something you enjoy (though it could be sickening).

  4. Driverless cars do bring some convenience to people, but the hidden safety problems behind them cannot be ignored. The 2016 Tesla accident is a good example. The driver’s death was caused by the driver’s negligence and systematic misjudgment. Therefore, this concept is like a double-edged sword. Using it well can bring convenience to drivers and reduce driving fatigue. Too much dependence can lead to accidents.

  5. Driverless car (and now driverless truck) technical and road-test developments are moving ahead at a rapid pace and perhaps without sufficient controls in place. The first ‘at fault’ deaths have occurred since the writing of this piece. The regulatory frameworks in most territories are playing catch-up and the companies involved are using this lack of control to move ahead despite the unknowns.

  6. Is there not a third option? Why can’t the car try to avoid killing both sets of people?
    A moral human would try to avoid fatalities at all cost and so the car could also be programmed to do the same; if then there are fatalities, the scenario is the same as humans crashing as the car hasn’t decided to kill anyone.

    An alternative could be that once the car has calculated a likely accident, it gives manual control of the car to the driver.

    1. Well I guess the car would only choose to kill someone if there was no third option. For example, if we look to the trolley example there is no third option, the rails only give two options.

      Your point of what humans would do is a very valid one, the AV could be programmed to save lives as opposed to deciding who to kill.

      I think the final point would be a bit shocking, imagine driving along nicely and then being given control suddenly when you’re not ready with an accident imminent.

  7. The ethical theories of utilitarianism and consequentialism are aptly included in a relevant scenario such as whether autonomous vehicles should be allowed on the roads, which is also seen in other high risk disciplines such as healthcare. Would the benefits that automated vehicles bring outweigh even the potential cost of one life taken as a result of a complex inference system? The families of the deceased would beg to differ. Autopilot is a feature of current vehicles, but there is still a level of control attributed to the human driver; can we really allow cars to be fully automated whereby failsafes may not work and human input ignored in difficult, complex situations which require human judgement?

    1. Interesting viewpoint. The family of the deceased would no doubt feel negatively towards the technology however, we have to remember that based on statistics there would be much fewer fatalities if the technology was implemented.

  8. Good write-up and discussion. Driverless cars seem like a great idea, get in and go, get out relaxed and refreshed. But if I’ve got to sit there wondering when something is going to go wrong and I might have to grab the controls then wouldn’t it be more relaxing just to take the controls in the first place?

    1. This is a very good point. If the system doesn’t have the trust of it’s users then many of the benefits of the system are diminished. I believe that the technology will build up a reputation over time allowing people to trust them but initially there will be that edginess.

  9. Choosing who to kill is a horrible scenario to be in, but it’s a reality that the system may have to face.

    I don’t think it’s possible to have a single algorithm that can decide who to kill -surely if humans cannot even decide upon morality how can a machine?

    The core question that comes out from all of this is how can we possible develop a machine that incorporates objective morality when no such thing can exist?!

    What is it that makes something right or wrong?

    And if by some miracle that algorithm was developed surely it would have to be adjusted depending on its place of deployment – as the choice of who to kill would be different depending on the location.

    I’m sure Israelis would much rather run over a Palestinian child than an elderly (illegal) occupier.

    Will be interesting to see how the human race gets around these problems.

    Any ideas on when my Ford Focus will be able to turn into a robot?

    1. Very interesting viewpoint JuicyLucy77. It’s quite right that objective morality is not something which is agreed upon and will differ from place to place. In a situation where people are being killed this difference of opinion is a very serious issue.

  10. Believe it or not this is the future of transport. Or at least a huge part of it. Right now, we can see top tech companies like Google, Apple, Tesla, etc. competing for the top place in this field but I feel we may not see truly driverless cars on the roads for at least another 7 – 8 years. Therefore, it is essential that while focusing on the technological aspects of all this we focus on the ethical ones as well. It is important that we understand what we are working towards and what affect it will have on us humans.

  11. Are we fixing something that isn’t broke?
    What will happen when we end up in a world with autonomous cars and likely buses, trams… ? The amount of driving-related jobs that will be lost is high. Cars will likely only be the start. And what will happen for things like ambulances? Even if they do remain manual, general driving skill is likely to decline.
    Overtime individuals will end up being extremely poor drivers and what happens in the instances that a manual driver is needed – poor driving.
    Perhaps the issue of accidents due to manual cars is something we must accept. The majority of manual car-related accidents is due to human mistakes, including driving under the influence, driving when tired, texting at the wheel. Maybe we should reflect on what causes these manual accidents instead of looking to shove the blame onto something automatic.

    1. This is an angle that hasn’t been mentioned so far. It’s very true that jobs will be lost in some areas in public transport but many jobs will be created in the development of AV technology plus these jobs will be higher skilled. The loss of skill level is a very good point, when the driver has to take over will he have the skill level to make things right? Very interesting point of view.

  12. This is a very interesting topic to think about. The future of artificial intelligence and driverless technologies are something that will affect all of us. But the important thing to remember is no matter how much tmadvanced these technologies get, we must not completely rely on them. I feel that in the future humans will have to be even more alert than they are now while. Imagine a scenario where the car is on driverless mode and it suddenly decides to kill you. What would the person in the car do? There are still a lot of things to think about.

    1. The mention about the car deciding to kill the passenger is a very important and valid point. It was not included in the article due to a the word limit however, there was a study done where they asked people if they would be in favour of the car deciding to kill the driver in choosing the ‘lesser of the evils’ and most people said they were as long as they weren’t the ones in the car. This shows the selfish mindset we have as human beings. I don’t think I’d feel comfortable sitting in a car that may make such a decision against my will.

  13. Sure driverless cars are going to make our lives more easier but at the same time they are going to create other problems as well. Think about what this would mean for the cybercrime world. Hackers would be able to gain control of the cars and could easily kill someone they wanted. Assassinations would take place and they would just look like normal car crashes due to system faults. People wouldn’t be safe, even in their own cars. Obviously with all this Cybersecurity would be on huge demand but still, no system is 100% secure especially if connected to the Internet. We need to decide if we are really ready to sacrifice all this just so we don’t have to drive. I mean surely it isn’t that bad now is it?

    1. Very true, even though AV’s are proven to be safer the system is more vulnerable to exploitation. I don’t imagine any high profile people will be allowed to use these vehicles so why should we feel safe? Some interesting things to think about.

  14. The implementation of AV’s on our roads is a difficult ethical consideration. A lot of variables that are not initially clear could occur. On one hand, if EVERY car was autonomous, then the safety benefits would be exceptional. However, are the controllers of legislation going to tell people they cant drive their own car? And what happens with emergency services transportation? As well as the hundreds of millions of non-autonomous vehicles that are already in existence? Av’s do show clear benefits to society, but either the technology would have to advance to a level that satisfies society, or society would have to accept to radical changes, for the full benefits to be appreciated.

    1. I think the changes will occur over time but eventually we would probably end up with a fully automated system for the greater good of society in the sense that AV’s are safer. The remark regarding emergency vehicles is interesting but I guess that if a car was smart enough to drive around then it could also get out of the way of emergency vehicles.

  15. This is really interesting. Where would blame lie if something were to go wrong? Would it lie with the engineer or developer, or then lie with the owner? Also, how would insurance work?

    1. My personal opinion is that the blame should lie with the tech companies if there is a one off fault but if there is something which was not considered then this would have to be with the regulators and the governments who did not put sufficient control in place to prevent the mishaps. I have no idea regarding insurance, I guess we’d need a whole new system altogether. Maybe a form of warranty from the AV company that if there is an accident they will pay.

  16. Always an interesting discussion. If robots can flawlessly carry out mathematical and data analyses, couldnt they also find the right answers to life? What happens if, perhaps, the answers they reach oppose the right to human life?

  17. An interesting read about the AV’s but one thing that keeps me thinking about this issue is it’s susceptibility to hacking or unauthorised controlling. In this modern era, no one will deny as to how easy it is for hackers to gain access to highly confidential files or documents. No wonder, this is just an AV.
    Well, the best-case scenario is that hackers take control safely but demand money to give it back -what’s called “ransomware attacks” but what’s more concerning would be taking over cars and using this new technology for filling cars with explosives, then having them drive to a destination with no one inside. Also, the police department of various countries have already studied how criminals could become more dangerous once they have the ability to shoot at police during car chases without having to watch the road.
    All these aspects definitely pushes us to think “Do we really need AVs?”

    1. The comment which really resonated with me was the use of AV’s to transport weapons and other illegal materials. The cars would need to be fitted with a sophisticated system which would prevent the car from being driven without someone inside although it would be really handy to be able to send the car off to deliver things. These things would need to be weighed up and addressed.

  18. Good write-up and discussion. Driverless cars seem like a great idea, get in and go, get out relaxed and refreshed. But if I’ve got to sit there wondering when something is going to go wrong and I might have to grab the controls then wouldn’t it be more relaxing just to take the controls in the first place?

  19. The pros and cons of artificial intelligence, a valuablt problem which is worthy of thinking. Definitely, technological process is a double-edged sword, but we are used to
    ignoring the nefative effects because of convinience from it. In my view, it will be terrible if we live in a word controlled by robots. Who can protect the rights of the general public once an accident happens like mentioned in the blog ?

    1. Ultimately the humans will ensure that they are still in control however, the developments in AV highlight the advancements in AI that are taking place. If the robots can work for us then this would make things much easier for us but as some of the greatest scientists have highlighted, AI could be our own downfall.

  20. The core question that comes out from all of this is how can we possible develop a machine that incorporates objective morality when no such thing can exist. But the important thing to remember is no matter how much tmadvanced these technologies get, we must not completely rely on them.

  21. Av’s do show clear benefits to society, but either the technology would have to advance to a level that satisfies society, or society would have to accept to radical changes, for the full benefits to be appreciated.

  22. This may sound very negative but I don’t understand why we always feel the need to keep changing things. The road system we have now works just fine and by introducing AV’s we are just creating more problems for ourselves.

  23. I think that there are good arguments on both side. Overall, I believe that there are more benefits that downsides although I do agree that if the system is proved unsafe then I wouldn’t want to be in one.

  24. I found the trolley thought experiment really interesting. I think that choosing who gets killed is very unethical, how do we value one human life to another? A better option would be to make the car save all lives and if in the process the car kills people then at least the best effort was made. I would have been pro AV’s all the way before but this article showed me that there are still many different things that we need to think about before we allow these vehicles to become mainstream.

  25. Great read. I think the thing that gets to me is the ways that this technology could be exploited by people who don’t have good intentions. I imagine that high value targets such as the leaders of countries won’t be allowed to use these kinds of vehicles in case of assassination attempts which just highlights the vulnerability of anything which is connected to the internet. I won’t be in any hurry to jump in one of these cars.

  26. I’m not a fan of robot developments, I feel like we’re sprinting forwards into the dark with our eyes closed. I believe that the development of automated systems is unethical because it may ultimately lead to the end of the human race. It amazes me how these tech companies can be allowed to just develop freely at will without any strict regulations, seems like a recipe for disaster to me.

  27. Interesting topic and amazing article. I think there is a huge potential in this field and we need to think about how we can improve it in such a way that we do not have any drawbacks.

  28. Good article. We need to think about what this technology will change. I mean, as mentioned above, if the driver-less car has the ability to kill its passengers then you really wouldn’t want children sitting in that car. In fact, you wouldn’t want anyone sitting in that car.

  29. Actually, I imagine that high value targets such as the leaders of countries won’t be allowed to use these kinds of vehicles in case of assassination attempts which just highlights the vulnerability of anything which is connected to the internet. And I think that choosing who gets killed is very unethical, how do we value one human life to another? A better option would be to make the car save all lives and if in the process the car kills people then at least the best effort was made.

  30. The development of robots is a trend of this era and a future development trend. This is something we cannot avoid. More and more robots now replace what should have been done by humans, especially those that do not require technical knowledge. However, people’s concerns about the development of robots are not unreasonable. Many movies also reflect the problems that may be caused by over-development of robots. Therefore, I think that the development of technology can be, but this technology must be considered to be able to control.

  31. Ultimately the humans will ensure that they are still in control however, the developments in AV highlight the advancements in AI that are taking place. It is true. But hackers would be able to gain control of the cars and could easily kill someone they wanted. That’s a problem.

  32. The argument of this article is very interesting. Technology will change people’s lives. Automated driving will also change people’s driving habits. However, if the safety of automatic driving is not guaranteed, the harm caused by it will be far greater than that of human-controlled vehicles.

  33. The autonomous car would be increasingly popular in future, but the safety must be the most controversial problem. A better option would be to make the car save all lives and if in the process the car kills people then at least the best effort was made.

  34. It is essential that while focusing on the technological aspects of all this we focus on the ethical ones as well. It is important that we understand what we are working towards and what affect it will have on us humans. Definitely, technological process is a double-edged sword, but we are used to ignoring the nefative effects because of convinience from it.

  35. I think that choosing who gets killed is very unethical, how do we value one human life to another? A better option would be to make the car save all lives and if in the process the car kills people then at least the best effort was made. Protection of personal information and security are very real concerns which need answering, who’s responsibility will it be to counter cyber threats to AV’s? – no doubt it will be a costly undertaking!

  36. A lot of variables that are not initially clear could occur. On one hand, if EVERY car was autonomous, then the safety benefits would be exceptional. However, are the controllers of legislation going to tell people they cant drive their own car? And what happens with emergency services transportation? A better option would be to make the car save all lives and if in the process the car kills people then at least the best effort was made.

  37. Self-driving cars would be very much vulnerable to cyber-attacks by hackers, though only a film concept, there are real safety concerns around this issue. And social contract theory also states a type of rule-ethical egoism stating that all our actions are selfishly motivated, people are better off living in a world with moral rules than one without moral rules. It outlines how we are subject to the whims of other people’s selfish interests in the absence of moral rules, in this case, the hackers. I think protect personal information is very important.

  38. I also think spam jams and hacker-driven congestion may influence self-driving experience in the future. If they control thousands of self-driving cars, the whole city will be in messy which is a horrible thing. We should do something to protect our personal information.

  39. Network data security is very important for ordinary citizens, especially government officials. How to protect personal privacy and information security are big problem. In this case, relevant rules and laws are important to be enacted and they are not perfect now.

  40. If the autonomous car kills somebody, it always makes people feel like that technology kill somebody instead of an accident. It’s hard to make people change this idea. The driver’s death was caused by the driver’s negligence and systematic misjudgment. Therefore, this concept is like a double-edged sword.

  41. Hackers may hijack vehicles for remote manipulation or even using them for criminal purposes as weapons. Self-driving cars may be in danger from hackers. How to protect personal privacy and information security are big problem. Imagine a scenario where the car is on driverless mode and it suddenly decides to kill you. What would the person in the car do? It is horrible.

Leave a Reply