Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Need for speed? Why vehicles capable of driving faster than legal speed limits should be banned

Need for speed? Why vehicles capable of driving faster than legal speed limits should be banned Speeding is a major cause of avoidable deaths and serious injuries. In this article, we defend the view that, with few excep- tions, vehicles should be required by law to have a limited intelligent speed assistant (LISA) fitted, making it impossible to exceed speed limits. Our core argument appeals to the four-element Principle of Required Prohibition: if (1) through suit- able legal regulation of design and production, the state can significantly reduce the harm users of a product P do to others through P’s illegal use without thereby causing comparable harms resulting from people not breaking the law; (2) that the relevant use of P ought to be illegal; (3) the regulation does not make any significant lawful use of P no longer possible; and (4) there is no other feasible and more efficient way of avoiding the relevant harm that users of P do others through P’s illegal use, then the state ought to regulate the design and production of P to significantly reduce the harm users of P do others through P’s illegal use. This principle, we argue, is extremely weak and should be acceptable across a wide range of ethical theories, yet it implies that the fitting of LISA to vehicles (but not police cars, ambulances etc.) should be mandatory by law. We defend the principle against five possible objections. For example, that the Principle of Required Prohibition is too interventionist and anti-libertarian. To counter that worry, we argue that principles that are even more interventionist are justifiable. Our conclusion is that the state should make it impossible for drivers to illegally violate speed limitations by making the installation of LISA on cars compulsory by law, just as the state, in many other ways, renders it impossible, or at any rate harder, for us to violate laws. Keywords AI and driving · Ethics of legal regulation · Limited intelligent speed assistant (LISA) · Speed limits · Risk imposition · Traffic ever produced. However, in Italy and many other counties, 1 Introduction the legal speed limit on motorways is 130 km/h (80 mph). It is considerably lower on other roads. In urban areas, the With a top speed of 410 km/h (254 mph), the Italian Bugatti limit is usually around 50 km/h (30 mph). But it is not only Veyron Grand Sport Vitesse is one of the fastest roadsters sport cars from famous car companies, such as Bugatti, Ferrari, Porsche, Chevrolet and Kia, which are designed to drive considerably faster than the maximum speed that most countries permit. Almost all cars can exceed the speed limit. The authors of this article both own a car, neither of which is particularly fancy, but both of which have a top speed of approximately 160–90 km/h which is 30–60 km/h more than * Thomas Søbirk Petersen the maximum speed limit in our country. thomassp@ruc.dk Kasper Lippert-Rasmussen lippert@ps.au.dk https:// www. bug at ti. com/ models/ v e yr on- models/ tec hn ology/ (accessed April 12, 2023). Department of Philosophy and Science Studies, Roskilde Germany and parts of Australia (in the Northern Territories) are University, P.O. Box 260, 4000 Roskilde, Denmark the only places in the world where, on certain highways, there are no Department of Political Science, Aarhus University, speed limits. However, even in these areas there are speed limits on Aarhus C, Denmark most roads. Vol.:(0123456789) 1 3 AI and Ethics Speeding is one of the main causes of traffic accidents meaning that when drivers depress the accelerator pedal, [7], p. 229, [6], [26], pp. 391–458, [24], 509). According to they cannot increase the vehicle’s speed beyond the maxi- a report from WHO, in 2018, there were around 1.3 million mum limit. Drivers cannot override this form of intelligent road deaths, and between 20 and 50 million people were assistance. Driving with LISA implemented in your car is injured (non-fatal) as result of road accidents involving all compatible with being, in all respects other than choos- type of vehicles, cyclists, and pedestrians [30]. Traffic acci- ing to speed, fully in charge of how you drive. The driver dents are “the leading cause of death for children and young still steers, applies the brakes, controls the lights, and so adults aged 5–29 years” [30], p. 1. A study from the Euro- on. The only limitation is that the driver cannot exceed the pean Transport Safety Council (ETSC) shows that 2100 lives legal speed limit even if the car’s engine would make this could be saved every year in EU countries “if the average option available in the absence of LISA. In Sect. 2, we shall speed dropped by 1 km/h” [10]. Another study of almost consider alternative strategies to limit speeding, explaining 30 European countries shows that approximately 50% of all how they work. traffic has drivers exceeding the legal speed limit in it [8 ]. The only publication addressing our question that we Facts such as these form the background to an ongo- have been able to find is an article by Jilles Smids [27]. ing and nuanced discussion among applied ethicists about Smids defends a conclusion like ours, but it presents a dif- where to set legal speed limits [18, 25, 31]. One question ferent argument appealing directly to the value of reducing is whether, and if so how, speed limits should vary across the number of traffic victims. Our argument appeals only different vehicles, including autonomous/semi-autonomous indirectly to this feature. Primarily, it relies on what we call vehicles [22], non-autonomous vehicles [2], electric scooters the Principle of Required Prohibition (to be explained in [13], etc. Another is when and indeed whether, driving above Sect. 3). One important advantage of this principle is that it the speed limit is morally wrong (e.g., [14, 19, 21]. These is extremely weak and therefore, we argue, acceptable across questions are important, but addressing them should not dis- a wide range of ethical theories. tract us from a third, and in our view, even more important There are several less important concerns supporting question: Do we have a moral obligation to ban the produc- mandatory installation of LISA. Here we simply want to list tion, sale and purchase of vehicles capable of driving faster these. First, the avoidance of accidents aside, drivers will than the legal limit? If motorists never took advantage of benefit if LISA is fitted to their vehicles by receiving fewer their vehicles’ capacity to travel faster than the law permits, speeding tickets and reduced insurance premiums [1]. Sec- perhaps the answer would be No. However, this is not the ond, it will become easier for drivers to comply with traffic situation. As the statistics above indicate (and many read- law. As a result, they will be fined and jailed less (which ers might have first-hand experience of motorists who drive presumably we can all agree is a good thing when it reflects faster than is legally permitted), drivers often speed. Again, a reduced number of offenses). The costs to the state of traf- as mentioned above, speeding causes thousands of people to fic law enforcement and the associated penal system will be avoidably killed, and many more to be avoidably injured, also be lower. Third, intelligent speed adaption will in gen- every year. In light of this, we shall argue that, with some eral reduce speeding and speed variance. This will result in common-sense exceptions, vehicles should be required by law to have a Limited Intelligent Speed Assistant (LISA), which prevents speeding, fitted. In a LISA-enabled car, it Smids [27] also raises sensible doubts about whether imposition of is impossible to exceed the official speed limit. The system the risk on others that is involved in driving within the speed limits is works by limiting the amount of fuel the engine can use, morally permissible. For present purposes, we ignore this issue and focus on illegal speeding. ETSC [9] points to the way LISA would benefit drivers by reducing the extent to which they are fined for speeding. Admittedly, law-abid- 3 ing drivers might deny that this gives them any reason for preferring In Great Britain in the year ending June 2022 approximately, 1,800 LISA, and those who ignore speed restrictions might reject the argu- deaths and serious injuries to 28,000 individuals were attributed to ment on the grounds that it is paternalistic (“With what right does the road accidents (National statistics 2022). state prevent us from owning cars that can go faster than the speed Strictly, this fact does not entail that stopping cars driving faster limit permits on the basis that this is better for us, when we ourselves than the legal speed limit would reduce the number of traffic victims. want to own such cars?”). In response to the first point, one might ask Logically, it is possible that all traffic victims were victims of acci- what reasons law-abiding citizens have to oppose mandatory imple- dents involving cars that were not exceeding the legal speed limits. mentation of LISA. In response to the second, it might be noted that However, it is extremely improbable that this is the case. most people accept some paternalistic legal requirements on the con- At least for the use of vehicles on public roads. Racing on tracks struction of cars—most notably, mandatory fitting of seat belts. In the and on private property is not our concern. relation to both points (assuming one grants them), one might also We say “virtually all” because we set aside issues raised by the per- ask whether they have sufficient moral weight to outweigh the moral missible speed capacities of emergency vehicles such as police cars, concern to prevent thousands of avoidable annual deaths and injuries. ambulances, fire engines, etc. 1 3 AI and Ethics more homogeneous traffic flow, fewer queues, less stress for drawback since speeding in such areas involves greater risks road users [12] and cuts in CO emissions [23, 29]. Fourth, than speeding on the highways [6]. crimes in which fast cars are used to escape from the police There is an alternative strategy to MSS, the Flexible would become more difficult to engage in (assuming that, as Maximum Speed Strategy (FMSS). FMSS involves fitting an exception, police cars are not implemented with LISA and an Intelligent Speed Assistant (ISA) in all vehicles. ISAs that criminals use cars with working LISA on them). Fifth, use GPS data and/or traffic-sign-recognition cameras to presumably everyone would agree that driverless cars should determine the maximum speed allowed in an area. They be designed so that they cannot go faster than the speed come in many versions. The most common are the Advisory limit. But if that is so, why should the same restriction not Intelligent Speed Assistant (AISA) and the Limiting Intel- 9 11 apply to cars with drivers? Finally, if LISA is implemented, ligent Speed Assistant (LISA). AISA sends haptic, audio some of the resources now being devoted to limiting speed- and visual warnings until you start driving within the speed ing, such as putting up speed bumps, could be redirected to limit, but it is still up to the driver to switch it on and comply making traffic safer in other ways. with the AISA advice that the car’s speed should be reduced. Section 2 briey fl describes the technologies of speed limi - From 2024, AISA is mandatory in all new cars and trucks in tation and how effective they are in reducing the number of EU [11]. LISA is more proactive. Once it is fitted, the driver accidents. Second 3 sets out our core argument for manda- cannot overrule the advice or warnings given by ISA. The tory LISA fitting. Section  4 responds to five objections. Sec- ISA system limits the engine’s power and thus the vehicle’s tion 5 briefly concludes. speed to ensure the legal speed limit is not breached. Stud- ies have shown that LISA is between three and ten times more effective in reducing accidents than AISA [5 , 17, 27]. 2 Two strategies to limit speeding It hardly needs adding that this is a significant difference. There are several strategies for combatting speeding. These include police control on the roads, speeding cameras, warn- 3 T he principle of required prohibition, ing signs, speed bumps, road or lane narrowing, the educa- LISA, and assault rifles tion of drivers, etc. However, despite many efforts of these kinds, people still speed. Therefore, more direct strategies to Making LISA legally mandatory in cars will save lives and combat speeding are surely worth looking at. The most obvi- reduce injuries in sustained road traffic [11, 27]. Experimen- ous strategy—call it the Maximum Speed Strategy (MSS)— tal trials and field tests show that fitting LISA in all vehicles is to regulate the design and production of vehicles in such a would reduce the number of fatal accidents by 25–50% (e.g., way that ordinary cars can no longer exceed a certain speed [5, 11, 20]). Given that approximately 30,000 people die limit, such as 130 km/h. This strategy is already used in in road accidents each year in the EU alone, this suggests, connection with 50 cc mopeds and the electric scooters that everything else being equal, that between 7500 and 15,000 are available to rent in many cities around the world. The lives could be saved each year if LISA were to be fitted on engines of such vehicles are very often designed to ensure all cars in the EU [11]. In addition, the same studies referred that they cannot drive faster than law permits (for electronic to above show that general implementation of LISA would scooters the speed limit is often 20  km/h, and for 50  cc reduce the number of people who are injured by road acci- mopeds it is 45 km/h). Applying a similar speed restriction dents by around 30%. Given that approximately 1.3 mil- to cars would have clear benefits in terms of improving traf- lion people in EU are injured each year (2017), this—again, fic safety. However, this strategy also has clear limitations. everything else being equal—is equivalent to a decrease in Most importantly, it would not prevent car drivers from the number of people injured by road accidents of almost exceeding the speed limit in urban areas. This is a serious 400,000 a year in the EU alone. Clearly, these facts are highly relevant to questions about state regulation of the production, sale, and possession of cars. More generally, we submit that: It might be replied that the crucial difference is that drivers have a choice, whereas driverless cars do not. However, driverless cars could easily be modified to embody this feature: they could be equipped with a disregard-the-speed-limit button that passengers could press if they wished to. We assume no one would be in favor of this design feature in driverless cars. For a description of how Intelligent Speed Assistance works, Some car manufacturing companies do limit their vehicles. including LISA (also called ‘Speed Control Function’ in the follow- In Japan, for example, domestic market vehicles are limited to ing reference) (and it does), see https:// www. makeu seof. com/ what- 180  km/h—see e.g., https:// garag edrea ms. net/ car- facts/ why- are- japan is- intel ligent- speed- assist- isa- how- does- it- work/ (accessed April 12, ese- cars- limit ed- to- 112mph- 180kmh (accessed April 12, 2023). 2023). 1 3 AI and Ethics The Principle of Required Prohibition: If (1) through of cutting speeding. However, we can confidently say that suitable legal regulation of design and production, the current alternative methods, such as speeding cameras, the state can significantly reduce the harm users of warning signs, and driver education, have had a limited a product P do to others through P’s illegal use with- effect, leaving us with hundreds of thousands of annual vic- out thereby causing comparable harms resulting from tims of illegal speeding worldwide. people not breaking the law ; (2) the relevant use of Since the Principle of Required Prohibition might seem P ought to be illegal; (3) the regulation does not make aggressively interventionist and anti-libertarian, it is worth any significant lawful use of P no longer possible; and noting just how weak (and hence plausible) it actually is. In (4) there is no other feasible and more efficient way fact, we think principles that are even more interventionist of avoiding the relevant harm that users of P do oth- are justifiable, but to defend compulsory LISA fitting in all ers through P’s illegal use; then (5) the state ought to vehicles we have no need of these stronger principles. We regulate the design and production of P to significantly should stress that our principle applies to harms resulting reduce the harm users of P do others through P’s ille- from uses of a product that are illegal and ought to be so— gal use. not harms resulting from the uses that either are or should be legal. It also applies to harm users of the product impose Given that P stands for vehicles, an argument with the on others, not harms they impose on themselves. Finally, Principle of Required Prohibition as its main premise and it applies where regulation would not make any significant the claim that compulsory use of LISA in vehicles satisfies lawful use of the relevant product no longer possible, and conditions (1–4) give us the conclusion that the fitting of where alternative harm-reducing regulation regimes are not LISA in all vehicles should be required by law. feasible. We have already offered several reasons why (1) is satis- This means that, arguably at least, the Principle of ed fi in the case of LISA—most importantly, the compulsory Required Prohibition does not imply that cigarette and alco- fitting of LISA in all vehicles would significantly reduce the hol production should be banned. Not only does (2) appear number of fatal and non-fatal traffic accident victims. In the to be unsatisfied where those products are concerned, since next section, we shall address a couple of challenges to the unlike driving at 130 km/h in cities, it is unclear that con- idea that (1) is satisfied by LISA. We have not defended (2), suming a glass of wine with friends should be illegal, but which where LISA is concerned is the claim that, morally (3) clearly is not met. Banning cigarettes and alcohol would speaking, there ought to be speed limits (e.g., that it should negatively affect many forms of relaxation and socializing, not be permissible to drive 150 km/h in cities, thereby pos- even if it would also result in a significant health improve- ing a grave risk to fellow motorists as well as cyclists, pedes- ment for the population. It could be argued that the Principle trians, or simply bystanders). We think this is a claim that all of Required Prohibition does not even justify prohibition or most of our readers will grant us. At any rate, we know of the manufacture of assault rifles for sale to the public. It of no country that imposes no speed limits on driving, and must be acknowledged that some will argue that while using we are not aware of a political party—not even a stridently an assault rifle to shoot at innocent people is and should libertarian one—or an interest organization that promotes be illegal, the possession of such a firearm, and its use to the cause of speed-unlimited driving. Of course, one can protect oneself from culpable aggressors, should be legal. discuss what the speed limits should be, but this discussion Forbidding the manufacture and sale of assault rifles would is orthogonal to our argument, which can be presented on eliminate this significant and (in many US states) lawful the basis of the simple assumption that the morally required use of assault rifles in such a manner that (3) is arguably not speed limits are those our interlocutor believes to be mor- satisfied in the case of such rifles. But now consider: ally required. Claim (3) we defend below. As regards (4), where LISA Limiting Targets Assistant: AI engineers develop an is concerned, it is hard to be certain that there is no other inexpensive mini-computer that can easily be fitted to feasible and more efficient way of achieving a reduction of all assault rifles. The computer blocks the rifle when the relevant harm drivers impose on others through illegal the shooter wants to fire it at an innocent person, and use of their vehicles. If for no other reason, this is because more widely when the rifle is used for another illegal we may not yet have thought through all the alternative ways One such stronger alternative principle would replace 1) as fol- We are not implying that harms to the users of the product them- lows: “If 1*) through suitable legal regulation of design and pro- selves are morally unimportant. However, banning a product, or duction the state can significantly reduce the harm users of a prod- features of it, out of concern for the users of the product themselves uct P do (both to the user and others) through P’s illegal use without invites anti-paternalistic argumentation, and we want to sidestep this thereby causing comparable harms resulting from people not break- here. ing the law.”. 1 3 AI and Ethics purpose, but it does not block the rifle from shooting and drive smart LISA cars that would prevent them from when the target is a legal one, as it would be when a speeding, limiting the risks they pose to others? culpable aggressor is about to attack the person with Admittedly, there are differences between the two cases. the assault rifle life-threateningly. For one thing, firing an assault rifle at an innocent person comes with a high likelihood of causing death and (if they Of course, there are no such smart assault rifles on the are lucky) less than fatal harm to them. Relatively speaking, market. The point we want to make by presenting Limit- driving over the legal limit is less likely to result in death or ing Targets Assistant (LTA) is that if such smart assault injury to innocent bystanders (it depends of course on how rifles could be produced and sold, then according to The fast you drive, and where). However, we deny that these Principle of Required Prohibition—and in our view very differences justify treating the two cases differently. Specifi- plausibly so—the state should outlaw production of con- cally, regarding the difference just mentioned, we would like ventional assault rifles which, unlike LTA rifles, can be used to point out that, intuitively, we would not think differently for various illegal purposes. Admittedly, those who want about the regulation of assault rifles if the risk of death and an assault rifle that can be used for illegal purposes but have injury from their use was lower—e.g., because the world in no intention of acting illegally might complain that ban- question is one in which people wear protective Kevlar gear. ning conventional assault rifles frustrates their strong desire Moreover, there are countervailing factors favoring LISA to possess such a rifle. However, that complaint does not cars more strongly than LTA assault rifles. Most importantly, seem to have much moral weight. First, one might question motorists very often speed, whereas rifle owners only very whether a desire to be able to do something that is, and rarely fire them at others, and this means that, from a purely should be, illegal carries any moral weight at all. Second, pragmatic point of view, speeding is a much more urgent even if we agree that the desire carries some moral weight, regulatory issue than shooting, other things been equal. it seems that in comparison with the concern to avoid people being subjected to death and serious injury its weight is lim- ited. Surely, few of us would think that my interest in having 4 T he fast and the suspicious: arguments a lab capable of producing weapons-grade anthrax (and in against LISA the absence of my actually producing any such anthrax) is sufficiently significant for it to outweigh the public interest The previous section contains our positive argument for the in banning the establishment and ownership of such labs. conclusion that LISA should be mandatory in law. In this If these two points are accepted, the question becomes: section, we respond to five challenges to this conclusion. In what way, if any, is our hypothetical LTA technology dif- First, some have argued that LISA would introduce novel ferent from the actual LISA technology? If we agree that risks, as we sometimes need to exceed the speed limit in something like The Principle of Required Prohibition shows ways that increase road safety (e.g., [22]. Overtaking, for that, and helps to explain why, the state should not allow the example, would become more dangerous, as the technology manufacture and sale to the public of conventional assault may prevent the driver from accelerating to a sufficiently rifles, when instead people could buy, own and use smart high speed to overtake a car without running into oncoming LTA assault rifles, should we not similarly agree that the drivers [15]. Therefore, mandatory fitting of LISA does not state should not allow the production of cars that can be used satisfy (1) in the Principle of Required Regulation. to exceed speed limits, creating an excessive risk of death or This is a real concern, but we think it has insufficient injury to other people, when instead people could buy, own weight to overturn our argument. When introducing a new technology, we cannot know for sure what will happen, but this should not make us stop the introduction of new tech- Probably, pragmatic concerns about LTA assault rifles would nology if we have strong reasons to believe that the risks arise—e.g., could the state manipulate the LTA technology for invidi- of negative outcomes are by far outweighed by the positive ous political purposes? We ask the reader to imagine all such con- outcomes. Safety vests on ferries introduced the risk that cerns have been satisfactorily addressed. None of this is to suggest that cars and assault rifles somehow fall in the same moral category (see also the two paragraphs below). We No doubt, there are such situations. However, the relevant question mention the case of assault rifles, first, to show how weak the Princi- is: Would drivers be sufficiently capable of identifying them correctly ple of Required prohibition is, and second, to support the claim that or would they incorrectly think, when overtaking, that increasing an interest in having the option to use a certain product in an illegal their speed enables them to avoid colliding with the oncoming vehi- way (without intending to do so) carries little moral weight when we cle? are deciding how to regulate the production and use of the relevant product. The case of assault rifles can be used for these purposes even For a splendid ‘all things considered defense’ of self-driving cars, if it is true that assault rifles and cars differ in that potential harm is even though they would not eradicate the existence of road accidents, a by-product of the latter but not of the former. We thank an anony- see [16]. Our proposal is much less radical, of course, than the pro- mous reviewer for pointing out the need for this clarification. posal to make self-driving cars mandatory. 1 3 AI and Ethics passengers would suffocate from misapplying them, e.g., ways that leave the option of harming people intact ([27], p. by getting the attached straps around their neck and the like. 215). Second, to the extent that mandatory implementation However, the fact that making safety vests mandatory on of LISA would involve a small cost in the form of limited ferries introduced this novel risk was (and is) not a good rea- erosion of the driver’s moral agency, this might be worth son on balance to decline to introduce them, since the risks it. Indeed, it would seem morally indecent to insist on the eliminated in this case—of drowning in an emergency—are avoidance of a small reduction of drivers’ moral agency at far graver than the novel risks introduced. Research since the cost of the deaths of thousands of people each week the 1990s has shown that the risks of associated with LISA ([27], p. 215). Similarly, it would be morally outrageous are very low, and that the rewards of requiring it to be fitted of the owners of assault rifles to resist, on grounds of the by law would clearly outweigh them ([27], p. 213). Fur- protection of their moral agency, mandatory fitting of Lim- thermore, car simulator experiments suggest that drivers in iting Targets Assistants on their assault rifles—e.g., in the cars fitted with LISA will tend to adjust their driving by presence of the infamous school shootings involving assault overtaking less (Jamson et al., p. 22). Supplementary and rifles with which we are all too familiar. Indeed, it seems mandatory programs educating drivers about when it is safe problematic to assign any moral weight at all to the preserva- to overtake in a car with the LISA system could be run (Jam- tion of an illegal option to engage in activities that are very son et al., p. 22). Finally, if LISA is combined with a system risky for other people. that “… would assist drivers in judging the safety of gaps in Brownsword’s argument also seems to show too much. the oncoming traffic stream, by means of a simple interface. Suppose cars could be redesigned so that they could not Such a support system could provide a valuable feedback be stopped by traffic barriers erected to protect pedestrians mechanism for drivers regarding the safety of overtaking (one sees these in cities now, around buildings vulnerable opportunities and, if combined with a mandatory ISA, could to terrorist attack). On Brownsword’s view, such redesign realize the traffic safety benefits reported elsewhere” ([12], would boost the moral agency of drivers, since it adds a mor- p. 22). ally unacceptable alternative to their set of options which, Second, Brownsword [3] worries that the fitting of LISA exercising moral judgment, they choose not to take. Clearly, to most vehicles would erode the moral agency and moral though, this would not justify redesigning cars in this way. responsibility of drivers. Indeed, he believes that what he There are good reasons why we want cars to continue to be calls “techno-regulation”, of which LISA is an example, in built in such a way that they can be stopped by traffic bar - designing out the option of agents inflicting harm on others, riers. The same is the case, we would argue, with building makes us passive moral agents “… that are no longer active cars so that they cannot travel faster than the speed limit. participants in a moral community” [3], p. 19). A third concern, also raised by Brownsword [3], is that In response to this, we note first that while LISA reduces the implementation of LISA for safety reasons is the first drivers’ moral agency in speeding, the moral agency of the step on a slippery slope that leads to a totalitarian society driver is still genuine. Drivers of LISA cars are, within the like the one described by George Orwell in his book 1984, legal speed limit, still in full control of overtaking, tail- in which the state uses technology extensively to regulate gating, and deciding to be or simply being inattentive in citizens’ behavior in detail. To our mind this worry about the potentially harmful effects of implementing mandatory LISA is speculative and, unlike the clear and immediate benefits of implementing Admittedly, none of this shows conclusively that LISA carries LISA, undocumented. It is important to remind ourselves no significant risks. However, in our view, in this context, as in the that the state already regulates products in all sorts of ways context of risk assessment of technologies, it is a positive, though not conclusive, reason to think that LISA does not introduce significant with an eye to preventing us from using them in ways that novel risks that studies to identify such risks have not yet identified cause harm to others. Typically, we are accustomed to these any. Should it turn out that overtaking with LISA is intolerably haz- regulated products. We embrace them and certainly do not ardous, a weaker form of LISA that would only prevent the car form worry that we are on a slippery slope leading to a totalitar- driving, say, 15% faster that the permitted speed, could reduce that hazard to tolerable levels. ian society. Few of us think that road barriers preventing Relative to the six Autonomy Levels (AL) for driverless vehicles, motorists from driving down a pedestrian street are a dan- drivers of LISA-enabled vehicles retain a high level of autonomy, as gerous concession potentially driving us in the direction of automated steering and acceleration is not part of LISA as we have illiberal societies like Nazi Germany, the USSR, etc. Finally, envisaged it. Actually, it is far from straightforward to place LISA it is worth keeping in mind that totalitarianism is first and on the United States Department of Transportation [28] six levels of autonomy. One could argue that LISA cars would sit at level AL 1, foremost about what the laws say (assuming there are laws) since those acting “… as the driver, are responsible for driving the and not about making citizens accept these laws. Indeed, vehicle. [However] when engaged, the system can perform either one could imagine raising the legal speeding limits, mak- steering OR acceleration/braking”. In a LISA vehicle, the system can ing installation of LISA mandatory, thus, as a consequence only be engaged in braking. 1 3 AI and Ethics reducing speeding and, thus, the number of traffic victims. LISA is enforced. On the latter issue, we note that, in many It strains credibility to think that this would put us on to a countries, by law cars need to be checked regularly by path toward a techno-totalitarian society. authorized mechanics, and that such checks could easily Fourth, mandatory fitting of LISA would make vehicles reveal if the LISA system has been adjusted. more expensive. Some drivers might even be unable to buy vehicles as a result. This, it might be argued, unjustly dis- advantages them relative to other drivers. Presently, fitting 5 Conclusion LISA hardware (including camera, brackets, trim and wir- ing) costs between €186–249 per vehicle. If LISA were We have argued that the speed of private vehicles—not to become mandatory, economies of scale would probably emergency vehicles (such as police cars, ambulances, and lower the costs of LISA. Moreover, the risks involved in fire engines) should be restricted by mandatory fitting of speeding are not equally distributed: disadvantaged people LISA. Our argument rests on a normative principle, the who cannot afford to buy a vehicle are subjected to them Principle of Required Prohibition, and on various empiri- without enjoying the benefits associated with its being per - cal premises regarding the effects of mandatory fitting of missible to impose this risk on others. Similarly, every fourth LISA and alternative means of reducing illegal speeding. traffic victim is a cyclist or a pedestrian. Yet they pose little We have argued that the Principle of Required Prohibition or no serious threat to others through their chosen means of is surprisingly weak in the sense of being very undemanding movement [27], 207–208). Hence, the appeal to fair distribu- and, thus, should be accepted by most if not all of us, and tion of benefits and burdens by those opposing mandatory we have appealed to various research reports to support the installation of LISA backfires: it also serves as an argument empirical claims we have made. Finally, we have rebutted against not addressing these unequal risks. Finally, manda- five objections to the legal prohibition of cars without LISA. tory installation would, as we pointed out above, result in Our conclusion is that the state should make it impossible a range of cost savings—e.g., on emergency services, hos- for drivers to illegally violate speed limitations by making pitals, insurance premiums. Surely, these savings could be the installation of LISA on cars compulsory by law, just as used to subsidize the roll out of LISA to the point that the the state, in many other ways, renders it impossible, or at present pragmatic concern is met. any rate harder, for us to violate laws. The case for manda- Fifth, and finally, some might contend that mandatory fit- tory LISA is, in all essential respects, just like the case for ting of LISA will not help much because drivers who want to road barriers that prevent motorists from driving on streets speed are likely to tinker with, or pay others to tinker with, reserved for pedestrians, or (to take a rather different issue) their speed limitation systems to enable them to drive faster the case for withholding taxes at source, thus depriving citi- than the speed limits permit—or make other adjustments if zens of the opportunity to under-declare their income and they are otherwise dissatisfied with how the system func- evade taxes they are required to pay by law. tions [4]. In support of this contention, they might point out Acknowledgements We would like to thank Jesper Ryberg, Sebastian how many (especially young) users of 50 cc mopeds have Jon Holmen, Sune Lægaard, Fatima Sabir, and two anonymous review- them altered so that they are capable of going faster than ers for very valuable comments. they were designed to go. Author contributions Both authors have contributed the same, con- If the mandatory fitting of LISA were to make no differ - cerning design, writing, review, etc. ence to speeding, the case for the scheme would fall apart. However, the antecedent here is very implausible. The great Funding Open access funding provided by Royal Danish Library. Funding was supported by Carlsbergfondet (CF21-0181) and The Dan- majority of 50 cc mopeds have not been tinkered with, and ish Research Foundation (DNFR-144). whereas “tuning” 50 cc mopeds involves a relatively simple mechanical intervention, tampering with LISA might well Availability of data and material Not applicable. be beyond the capabilities of all but a few trained mechan- ics with supplementary courses in electronics and software. Finally, how ineffective the LISA technology becomes is Admittedly, unscrupulous drivers could conceal their tamper- likely to depend on the fines for tampering with it, and more ing with LISA from the mechanics. The checking regime would be generally how stringently the prohibition on cars without imperfect in this respect. However, regular (annual?) adjustments turning LISA on and off are likely to be complicated and costly. How the state should implement obligatory installation of LISA on cars if that is made compulsory by law is an important topic which Based on individual costs extracted from calculations made in we have not dealt with in the paper. One model would require all 2012 (see https:// www. 20spl enty. org/ what_ is_ isa (accessed April 12, vehicles produced from, say, 2024 to have implemented LISA, as is 2023)). Admittedly, on top of the indicated costs comes the cost of the case with ASIA in UK and EU. Another model would require all the software for LISA. vehicles to have LISA implements at a certain date. 1 3 AI and Ethics 11. EU Regulation: /2144 https://eur -le x.eur opa.eu/ eli/ r eg/2019/ 2144/ Declarations oj. Accessed 21 Feb 2023 (2019) 12. Hegeman, G., Tapani, A., Hoogendoorn, S.: Overtaking assistant Conflict of interests There are no potential conflicts of interest to dis- assessment using traffic simulation. Transport. Res. Part C Emerg. close. Technol. 17(6), 617–630 (2009) 13. Hirst, D.: Regulating electric scooters (E-scooters). House of Ethics approval and consent to participate In the submitted article, Commons Library, London. Briefing Paper, 8958(10) (2021) there are no research involving human participants and/or animals and 14. Husak, D.: Vehicles and crashes: why is this moral issue over- therefore no informed consent from these non-existent research sub- looked? Soc. Theory Pract. 30(3), 351–370 (2004) ject—so not applicable. 15. Jamson, S., Chorlton, K., Carsten, O.: Could intelligent speed adaptation make overtaking unsafe? Accid. Anal. Prev. 48, 29–36 Consent for publication Both authors consent to the publication men- (2012) tioned above. 16. Kamm, F.M.: The use and abuse of the trolley problem: self- driving cars; medical treatments, and the distribution of harm. Open Access This article is licensed under a Creative Commons Attri- In: Liao, M.S. (ed.) Ethics of artificial intelligence, pp. 79–108. bution 4.0 International License, which permits use, sharing, adapta- Oxford University Press, Oxford (2020) tion, distribution and reproduction in any medium or format, as long 17. Lai, F., Carsten, O., Tate, F.: How much benefit does intelligent as you give appropriate credit to the original author(s) and the source, speed adaptation deliver: an analysis of its potential contribution provide a link to the Creative Commons licence, and indicate if changes to safety and environment. Accid. Anal. Prev. 48, 63–72 (2012) were made. The images or other third party material in this article are 18. Lee, Y.M., Chong, S.Y., Goonting, K., Sheppard, E.: The effect of included in the article's Creative Commons licence, unless indicated speed limit credibility on drivers’ speed choice. Transport. Res. F otherwise in a credit line to the material. If material is not included in Traffic Psychol. Behav. 45, 43–53 (2017) the article's Creative Commons licence and your intended use is not 19. Norcross, A.: Comparing harms: headaches and human lives. permitted by statutory regulation or exceeds the permitted use, you will Philos. Public Aff. 26(2), 135–167 (1997) need to obtain permission directly from the copyright holder. To view a 20. Oei, H.L., Polak, P.H.: Intelligent speed adaptation (ISA) and road copy of this licence, visit http://cr eativ ecommons. or g/licen ses/ b y/4.0/ . safety. IATSS Res. 26(2), 45–51 (2002) 21. Ori, M.: Why not road ethics? Theoria 86(3), 389–412 (2020) 22. Reed, N., Leiman, T., Palade, P., Martens, M., Kester, L.: Ethics of automated vehicles: breaking traffic rules for road safety. Ethics References Inf. Technol. 23(4), 777–789 (2021) 23. Regan, M.: Preventing traffic accidents by mobile phone users. 1. Agerholm, N., Tradisauskas, N., Harms, L., Lahrmann, H.: Pre- Med. J. Australia 185(11/12), 628 (2006) liminary results from the Danish ISA Project Spar på farten- 24. Sahebi, S., Nassiri, H., Van Wee, B., & Araghi, Y.: Incorporating Behavior (2007) car owner preferences for the introduction of economic incentives 2. Augeri, M.G., Cozzo, P., Greco, S.: Dominance-based rough for speed limit enforcement. Trans. Res. Part F: Traffic Psychol set approach: an application case study for setting speed limits Behav. 64, 509–521 (2019) for vehicles in speed controlled zones. Knowl. Based Syst. 89, 25. Shill, G. H.: Should law subsidize driving? NYUL Rev. 95, 498 288–300 (2015) (2020) 3. Brownsword, R.: Code, control, and choice: why East is East and 26. Shinar, D. (Ed.). Traffic safety and human behavior. Emerald West is West. Leg. Stud. 25(1), 1–21 (2005) Group Publishing (2017) 4. Brinkley, J., Dunbar, J., Smith, J.D., Gilbert, J.E.: A Usability 27. Smids, J.: The moral case for intelligent speed adaptation. J. Appl. evaluation of the BMW active cruise control system with “Stop Philos. 35(2), 205–221 (2018) and Go” function. In: Proceedings of the Human Factors and 28. United States Department of Transportation: Automated vehicle Ergonomics Society Annual Meeting, vol. 61(1), pp. 1536–1540. for safety. https:// www. nhtsa. gov/ techn ology- innov ation/ autom Sage CA: Los Angeles, CA: SAGE Publications (2017) ated- vehic les- safety. Accessed 21 Feb 2023 (2022) 5. Carsten, O.: Is intelligent speed adaptation ready for deployment? 29. Várhelyi, A., Hjälmdahl, M., Hydén, C., & Draskóczy, M.: Effects Accid. Anal. Prev. 48, 1–3 (2012) of an active accelerator pedal on driver behaviour and traffic safety 6. Elvik, R., Vadeby, A., Hels, T., van Schagen, I.: Updated estimates after long-term use in urban areas. Acc. Anal. Prevent. 36(5), of the relationship between speed and road safety at the aggregate 729–737 (2004) and individual levels. Accid. Anal. Prev. 123, 114–122 (2019) 30. WHO: Road traffic injuries. https://www .who. int/ ne ws-r oom/f act- 7. Elvik, R.: Speed limits, enforcement, and health consequences. sheets/ detail/ road- tra ffic- injur ies. Accessed 21 Feb 2023 (2022) Annu. Rev. Public Health 33, 225–238 (2012) 31. Yao, Y., Carsten, O., Hibberd, D.: An empirical approach to deter- 8. ETSC: Road safety target in sight: making up for lost time. https:// mining speed limit credibility. Transport. Res. F Traffic Psychol. e tsc. eu/ 4t h- annual- r oad- saf e ty- per f o r mance- index- pin- r epor t/. Behav. 63, 270–282 (2019) Accessed 21 Feb 2023 (2010) 9. ETSC: Briefing: intelligent speed assistance (ISA). https://e tsc.e u/ Publisher's Note Springer Nature remains neutral with regard to brie fing- intel ligent- speed- assis tance- isa/. Accessed 21 Feb 2023 jurisdictional claims in published maps and institutional affiliations. (2017) 10. ETSC: Reducing speeding in Europe https:// e tsc. eu/ r educ ing- speed ing- in- europe- pin- flash- 36/. Accessed 21 Feb 2023 (2019) 1 3 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png AI and Ethics Springer Journals

Need for speed? Why vehicles capable of driving faster than legal speed limits should be banned

Petersen, Thomas Søbirk; Lippert-Rasmussen, Kasper
AI and Ethics , Volume OnlineFirst – May 16, 2023
Download PDF
8 pages

Loading next page...
 
/lp/springer-journals/need-for-speed-why-vehicles-capable-of-driving-faster-than-legal-speed-M4kIyXf6MV

References (23)

Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2023
ISSN
2730-5953
eISSN
2730-5961
DOI
10.1007/s43681-023-00290-9
Publisher site
See Article on Publisher Site

Abstract

Speeding is a major cause of avoidable deaths and serious injuries. In this article, we defend the view that, with few excep- tions, vehicles should be required by law to have a limited intelligent speed assistant (LISA) fitted, making it impossible to exceed speed limits. Our core argument appeals to the four-element Principle of Required Prohibition: if (1) through suit- able legal regulation of design and production, the state can significantly reduce the harm users of a product P do to others through P’s illegal use without thereby causing comparable harms resulting from people not breaking the law; (2) that the relevant use of P ought to be illegal; (3) the regulation does not make any significant lawful use of P no longer possible; and (4) there is no other feasible and more efficient way of avoiding the relevant harm that users of P do others through P’s illegal use, then the state ought to regulate the design and production of P to significantly reduce the harm users of P do others through P’s illegal use. This principle, we argue, is extremely weak and should be acceptable across a wide range of ethical theories, yet it implies that the fitting of LISA to vehicles (but not police cars, ambulances etc.) should be mandatory by law. We defend the principle against five possible objections. For example, that the Principle of Required Prohibition is too interventionist and anti-libertarian. To counter that worry, we argue that principles that are even more interventionist are justifiable. Our conclusion is that the state should make it impossible for drivers to illegally violate speed limitations by making the installation of LISA on cars compulsory by law, just as the state, in many other ways, renders it impossible, or at any rate harder, for us to violate laws. Keywords AI and driving · Ethics of legal regulation · Limited intelligent speed assistant (LISA) · Speed limits · Risk imposition · Traffic ever produced. However, in Italy and many other counties, 1 Introduction the legal speed limit on motorways is 130 km/h (80 mph). It is considerably lower on other roads. In urban areas, the With a top speed of 410 km/h (254 mph), the Italian Bugatti limit is usually around 50 km/h (30 mph). But it is not only Veyron Grand Sport Vitesse is one of the fastest roadsters sport cars from famous car companies, such as Bugatti, Ferrari, Porsche, Chevrolet and Kia, which are designed to drive considerably faster than the maximum speed that most countries permit. Almost all cars can exceed the speed limit. The authors of this article both own a car, neither of which is particularly fancy, but both of which have a top speed of approximately 160–90 km/h which is 30–60 km/h more than * Thomas Søbirk Petersen the maximum speed limit in our country. thomassp@ruc.dk Kasper Lippert-Rasmussen lippert@ps.au.dk https:// www. bug at ti. com/ models/ v e yr on- models/ tec hn ology/ (accessed April 12, 2023). Department of Philosophy and Science Studies, Roskilde Germany and parts of Australia (in the Northern Territories) are University, P.O. Box 260, 4000 Roskilde, Denmark the only places in the world where, on certain highways, there are no Department of Political Science, Aarhus University, speed limits. However, even in these areas there are speed limits on Aarhus C, Denmark most roads. Vol.:(0123456789) 1 3 AI and Ethics Speeding is one of the main causes of traffic accidents meaning that when drivers depress the accelerator pedal, [7], p. 229, [6], [26], pp. 391–458, [24], 509). According to they cannot increase the vehicle’s speed beyond the maxi- a report from WHO, in 2018, there were around 1.3 million mum limit. Drivers cannot override this form of intelligent road deaths, and between 20 and 50 million people were assistance. Driving with LISA implemented in your car is injured (non-fatal) as result of road accidents involving all compatible with being, in all respects other than choos- type of vehicles, cyclists, and pedestrians [30]. Traffic acci- ing to speed, fully in charge of how you drive. The driver dents are “the leading cause of death for children and young still steers, applies the brakes, controls the lights, and so adults aged 5–29 years” [30], p. 1. A study from the Euro- on. The only limitation is that the driver cannot exceed the pean Transport Safety Council (ETSC) shows that 2100 lives legal speed limit even if the car’s engine would make this could be saved every year in EU countries “if the average option available in the absence of LISA. In Sect. 2, we shall speed dropped by 1 km/h” [10]. Another study of almost consider alternative strategies to limit speeding, explaining 30 European countries shows that approximately 50% of all how they work. traffic has drivers exceeding the legal speed limit in it [8 ]. The only publication addressing our question that we Facts such as these form the background to an ongo- have been able to find is an article by Jilles Smids [27]. ing and nuanced discussion among applied ethicists about Smids defends a conclusion like ours, but it presents a dif- where to set legal speed limits [18, 25, 31]. One question ferent argument appealing directly to the value of reducing is whether, and if so how, speed limits should vary across the number of traffic victims. Our argument appeals only different vehicles, including autonomous/semi-autonomous indirectly to this feature. Primarily, it relies on what we call vehicles [22], non-autonomous vehicles [2], electric scooters the Principle of Required Prohibition (to be explained in [13], etc. Another is when and indeed whether, driving above Sect. 3). One important advantage of this principle is that it the speed limit is morally wrong (e.g., [14, 19, 21]. These is extremely weak and therefore, we argue, acceptable across questions are important, but addressing them should not dis- a wide range of ethical theories. tract us from a third, and in our view, even more important There are several less important concerns supporting question: Do we have a moral obligation to ban the produc- mandatory installation of LISA. Here we simply want to list tion, sale and purchase of vehicles capable of driving faster these. First, the avoidance of accidents aside, drivers will than the legal limit? If motorists never took advantage of benefit if LISA is fitted to their vehicles by receiving fewer their vehicles’ capacity to travel faster than the law permits, speeding tickets and reduced insurance premiums [1]. Sec- perhaps the answer would be No. However, this is not the ond, it will become easier for drivers to comply with traffic situation. As the statistics above indicate (and many read- law. As a result, they will be fined and jailed less (which ers might have first-hand experience of motorists who drive presumably we can all agree is a good thing when it reflects faster than is legally permitted), drivers often speed. Again, a reduced number of offenses). The costs to the state of traf- as mentioned above, speeding causes thousands of people to fic law enforcement and the associated penal system will be avoidably killed, and many more to be avoidably injured, also be lower. Third, intelligent speed adaption will in gen- every year. In light of this, we shall argue that, with some eral reduce speeding and speed variance. This will result in common-sense exceptions, vehicles should be required by law to have a Limited Intelligent Speed Assistant (LISA), which prevents speeding, fitted. In a LISA-enabled car, it Smids [27] also raises sensible doubts about whether imposition of is impossible to exceed the official speed limit. The system the risk on others that is involved in driving within the speed limits is works by limiting the amount of fuel the engine can use, morally permissible. For present purposes, we ignore this issue and focus on illegal speeding. ETSC [9] points to the way LISA would benefit drivers by reducing the extent to which they are fined for speeding. Admittedly, law-abid- 3 ing drivers might deny that this gives them any reason for preferring In Great Britain in the year ending June 2022 approximately, 1,800 LISA, and those who ignore speed restrictions might reject the argu- deaths and serious injuries to 28,000 individuals were attributed to ment on the grounds that it is paternalistic (“With what right does the road accidents (National statistics 2022). state prevent us from owning cars that can go faster than the speed Strictly, this fact does not entail that stopping cars driving faster limit permits on the basis that this is better for us, when we ourselves than the legal speed limit would reduce the number of traffic victims. want to own such cars?”). In response to the first point, one might ask Logically, it is possible that all traffic victims were victims of acci- what reasons law-abiding citizens have to oppose mandatory imple- dents involving cars that were not exceeding the legal speed limits. mentation of LISA. In response to the second, it might be noted that However, it is extremely improbable that this is the case. most people accept some paternalistic legal requirements on the con- At least for the use of vehicles on public roads. Racing on tracks struction of cars—most notably, mandatory fitting of seat belts. In the and on private property is not our concern. relation to both points (assuming one grants them), one might also We say “virtually all” because we set aside issues raised by the per- ask whether they have sufficient moral weight to outweigh the moral missible speed capacities of emergency vehicles such as police cars, concern to prevent thousands of avoidable annual deaths and injuries. ambulances, fire engines, etc. 1 3 AI and Ethics more homogeneous traffic flow, fewer queues, less stress for drawback since speeding in such areas involves greater risks road users [12] and cuts in CO emissions [23, 29]. Fourth, than speeding on the highways [6]. crimes in which fast cars are used to escape from the police There is an alternative strategy to MSS, the Flexible would become more difficult to engage in (assuming that, as Maximum Speed Strategy (FMSS). FMSS involves fitting an exception, police cars are not implemented with LISA and an Intelligent Speed Assistant (ISA) in all vehicles. ISAs that criminals use cars with working LISA on them). Fifth, use GPS data and/or traffic-sign-recognition cameras to presumably everyone would agree that driverless cars should determine the maximum speed allowed in an area. They be designed so that they cannot go faster than the speed come in many versions. The most common are the Advisory limit. But if that is so, why should the same restriction not Intelligent Speed Assistant (AISA) and the Limiting Intel- 9 11 apply to cars with drivers? Finally, if LISA is implemented, ligent Speed Assistant (LISA). AISA sends haptic, audio some of the resources now being devoted to limiting speed- and visual warnings until you start driving within the speed ing, such as putting up speed bumps, could be redirected to limit, but it is still up to the driver to switch it on and comply making traffic safer in other ways. with the AISA advice that the car’s speed should be reduced. Section 2 briey fl describes the technologies of speed limi - From 2024, AISA is mandatory in all new cars and trucks in tation and how effective they are in reducing the number of EU [11]. LISA is more proactive. Once it is fitted, the driver accidents. Second 3 sets out our core argument for manda- cannot overrule the advice or warnings given by ISA. The tory LISA fitting. Section  4 responds to five objections. Sec- ISA system limits the engine’s power and thus the vehicle’s tion 5 briefly concludes. speed to ensure the legal speed limit is not breached. Stud- ies have shown that LISA is between three and ten times more effective in reducing accidents than AISA [5 , 17, 27]. 2 Two strategies to limit speeding It hardly needs adding that this is a significant difference. There are several strategies for combatting speeding. These include police control on the roads, speeding cameras, warn- 3 T he principle of required prohibition, ing signs, speed bumps, road or lane narrowing, the educa- LISA, and assault rifles tion of drivers, etc. However, despite many efforts of these kinds, people still speed. Therefore, more direct strategies to Making LISA legally mandatory in cars will save lives and combat speeding are surely worth looking at. The most obvi- reduce injuries in sustained road traffic [11, 27]. Experimen- ous strategy—call it the Maximum Speed Strategy (MSS)— tal trials and field tests show that fitting LISA in all vehicles is to regulate the design and production of vehicles in such a would reduce the number of fatal accidents by 25–50% (e.g., way that ordinary cars can no longer exceed a certain speed [5, 11, 20]). Given that approximately 30,000 people die limit, such as 130 km/h. This strategy is already used in in road accidents each year in the EU alone, this suggests, connection with 50 cc mopeds and the electric scooters that everything else being equal, that between 7500 and 15,000 are available to rent in many cities around the world. The lives could be saved each year if LISA were to be fitted on engines of such vehicles are very often designed to ensure all cars in the EU [11]. In addition, the same studies referred that they cannot drive faster than law permits (for electronic to above show that general implementation of LISA would scooters the speed limit is often 20  km/h, and for 50  cc reduce the number of people who are injured by road acci- mopeds it is 45 km/h). Applying a similar speed restriction dents by around 30%. Given that approximately 1.3 mil- to cars would have clear benefits in terms of improving traf- lion people in EU are injured each year (2017), this—again, fic safety. However, this strategy also has clear limitations. everything else being equal—is equivalent to a decrease in Most importantly, it would not prevent car drivers from the number of people injured by road accidents of almost exceeding the speed limit in urban areas. This is a serious 400,000 a year in the EU alone. Clearly, these facts are highly relevant to questions about state regulation of the production, sale, and possession of cars. More generally, we submit that: It might be replied that the crucial difference is that drivers have a choice, whereas driverless cars do not. However, driverless cars could easily be modified to embody this feature: they could be equipped with a disregard-the-speed-limit button that passengers could press if they wished to. We assume no one would be in favor of this design feature in driverless cars. For a description of how Intelligent Speed Assistance works, Some car manufacturing companies do limit their vehicles. including LISA (also called ‘Speed Control Function’ in the follow- In Japan, for example, domestic market vehicles are limited to ing reference) (and it does), see https:// www. makeu seof. com/ what- 180  km/h—see e.g., https:// garag edrea ms. net/ car- facts/ why- are- japan is- intel ligent- speed- assist- isa- how- does- it- work/ (accessed April 12, ese- cars- limit ed- to- 112mph- 180kmh (accessed April 12, 2023). 2023). 1 3 AI and Ethics The Principle of Required Prohibition: If (1) through of cutting speeding. However, we can confidently say that suitable legal regulation of design and production, the current alternative methods, such as speeding cameras, the state can significantly reduce the harm users of warning signs, and driver education, have had a limited a product P do to others through P’s illegal use with- effect, leaving us with hundreds of thousands of annual vic- out thereby causing comparable harms resulting from tims of illegal speeding worldwide. people not breaking the law ; (2) the relevant use of Since the Principle of Required Prohibition might seem P ought to be illegal; (3) the regulation does not make aggressively interventionist and anti-libertarian, it is worth any significant lawful use of P no longer possible; and noting just how weak (and hence plausible) it actually is. In (4) there is no other feasible and more efficient way fact, we think principles that are even more interventionist of avoiding the relevant harm that users of P do oth- are justifiable, but to defend compulsory LISA fitting in all ers through P’s illegal use; then (5) the state ought to vehicles we have no need of these stronger principles. We regulate the design and production of P to significantly should stress that our principle applies to harms resulting reduce the harm users of P do others through P’s ille- from uses of a product that are illegal and ought to be so— gal use. not harms resulting from the uses that either are or should be legal. It also applies to harm users of the product impose Given that P stands for vehicles, an argument with the on others, not harms they impose on themselves. Finally, Principle of Required Prohibition as its main premise and it applies where regulation would not make any significant the claim that compulsory use of LISA in vehicles satisfies lawful use of the relevant product no longer possible, and conditions (1–4) give us the conclusion that the fitting of where alternative harm-reducing regulation regimes are not LISA in all vehicles should be required by law. feasible. We have already offered several reasons why (1) is satis- This means that, arguably at least, the Principle of ed fi in the case of LISA—most importantly, the compulsory Required Prohibition does not imply that cigarette and alco- fitting of LISA in all vehicles would significantly reduce the hol production should be banned. Not only does (2) appear number of fatal and non-fatal traffic accident victims. In the to be unsatisfied where those products are concerned, since next section, we shall address a couple of challenges to the unlike driving at 130 km/h in cities, it is unclear that con- idea that (1) is satisfied by LISA. We have not defended (2), suming a glass of wine with friends should be illegal, but which where LISA is concerned is the claim that, morally (3) clearly is not met. Banning cigarettes and alcohol would speaking, there ought to be speed limits (e.g., that it should negatively affect many forms of relaxation and socializing, not be permissible to drive 150 km/h in cities, thereby pos- even if it would also result in a significant health improve- ing a grave risk to fellow motorists as well as cyclists, pedes- ment for the population. It could be argued that the Principle trians, or simply bystanders). We think this is a claim that all of Required Prohibition does not even justify prohibition or most of our readers will grant us. At any rate, we know of the manufacture of assault rifles for sale to the public. It of no country that imposes no speed limits on driving, and must be acknowledged that some will argue that while using we are not aware of a political party—not even a stridently an assault rifle to shoot at innocent people is and should libertarian one—or an interest organization that promotes be illegal, the possession of such a firearm, and its use to the cause of speed-unlimited driving. Of course, one can protect oneself from culpable aggressors, should be legal. discuss what the speed limits should be, but this discussion Forbidding the manufacture and sale of assault rifles would is orthogonal to our argument, which can be presented on eliminate this significant and (in many US states) lawful the basis of the simple assumption that the morally required use of assault rifles in such a manner that (3) is arguably not speed limits are those our interlocutor believes to be mor- satisfied in the case of such rifles. But now consider: ally required. Claim (3) we defend below. As regards (4), where LISA Limiting Targets Assistant: AI engineers develop an is concerned, it is hard to be certain that there is no other inexpensive mini-computer that can easily be fitted to feasible and more efficient way of achieving a reduction of all assault rifles. The computer blocks the rifle when the relevant harm drivers impose on others through illegal the shooter wants to fire it at an innocent person, and use of their vehicles. If for no other reason, this is because more widely when the rifle is used for another illegal we may not yet have thought through all the alternative ways One such stronger alternative principle would replace 1) as fol- We are not implying that harms to the users of the product them- lows: “If 1*) through suitable legal regulation of design and pro- selves are morally unimportant. However, banning a product, or duction the state can significantly reduce the harm users of a prod- features of it, out of concern for the users of the product themselves uct P do (both to the user and others) through P’s illegal use without invites anti-paternalistic argumentation, and we want to sidestep this thereby causing comparable harms resulting from people not break- here. ing the law.”. 1 3 AI and Ethics purpose, but it does not block the rifle from shooting and drive smart LISA cars that would prevent them from when the target is a legal one, as it would be when a speeding, limiting the risks they pose to others? culpable aggressor is about to attack the person with Admittedly, there are differences between the two cases. the assault rifle life-threateningly. For one thing, firing an assault rifle at an innocent person comes with a high likelihood of causing death and (if they Of course, there are no such smart assault rifles on the are lucky) less than fatal harm to them. Relatively speaking, market. The point we want to make by presenting Limit- driving over the legal limit is less likely to result in death or ing Targets Assistant (LTA) is that if such smart assault injury to innocent bystanders (it depends of course on how rifles could be produced and sold, then according to The fast you drive, and where). However, we deny that these Principle of Required Prohibition—and in our view very differences justify treating the two cases differently. Specifi- plausibly so—the state should outlaw production of con- cally, regarding the difference just mentioned, we would like ventional assault rifles which, unlike LTA rifles, can be used to point out that, intuitively, we would not think differently for various illegal purposes. Admittedly, those who want about the regulation of assault rifles if the risk of death and an assault rifle that can be used for illegal purposes but have injury from their use was lower—e.g., because the world in no intention of acting illegally might complain that ban- question is one in which people wear protective Kevlar gear. ning conventional assault rifles frustrates their strong desire Moreover, there are countervailing factors favoring LISA to possess such a rifle. However, that complaint does not cars more strongly than LTA assault rifles. Most importantly, seem to have much moral weight. First, one might question motorists very often speed, whereas rifle owners only very whether a desire to be able to do something that is, and rarely fire them at others, and this means that, from a purely should be, illegal carries any moral weight at all. Second, pragmatic point of view, speeding is a much more urgent even if we agree that the desire carries some moral weight, regulatory issue than shooting, other things been equal. it seems that in comparison with the concern to avoid people being subjected to death and serious injury its weight is lim- ited. Surely, few of us would think that my interest in having 4 T he fast and the suspicious: arguments a lab capable of producing weapons-grade anthrax (and in against LISA the absence of my actually producing any such anthrax) is sufficiently significant for it to outweigh the public interest The previous section contains our positive argument for the in banning the establishment and ownership of such labs. conclusion that LISA should be mandatory in law. In this If these two points are accepted, the question becomes: section, we respond to five challenges to this conclusion. In what way, if any, is our hypothetical LTA technology dif- First, some have argued that LISA would introduce novel ferent from the actual LISA technology? If we agree that risks, as we sometimes need to exceed the speed limit in something like The Principle of Required Prohibition shows ways that increase road safety (e.g., [22]. Overtaking, for that, and helps to explain why, the state should not allow the example, would become more dangerous, as the technology manufacture and sale to the public of conventional assault may prevent the driver from accelerating to a sufficiently rifles, when instead people could buy, own and use smart high speed to overtake a car without running into oncoming LTA assault rifles, should we not similarly agree that the drivers [15]. Therefore, mandatory fitting of LISA does not state should not allow the production of cars that can be used satisfy (1) in the Principle of Required Regulation. to exceed speed limits, creating an excessive risk of death or This is a real concern, but we think it has insufficient injury to other people, when instead people could buy, own weight to overturn our argument. When introducing a new technology, we cannot know for sure what will happen, but this should not make us stop the introduction of new tech- Probably, pragmatic concerns about LTA assault rifles would nology if we have strong reasons to believe that the risks arise—e.g., could the state manipulate the LTA technology for invidi- of negative outcomes are by far outweighed by the positive ous political purposes? We ask the reader to imagine all such con- outcomes. Safety vests on ferries introduced the risk that cerns have been satisfactorily addressed. None of this is to suggest that cars and assault rifles somehow fall in the same moral category (see also the two paragraphs below). We No doubt, there are such situations. However, the relevant question mention the case of assault rifles, first, to show how weak the Princi- is: Would drivers be sufficiently capable of identifying them correctly ple of Required prohibition is, and second, to support the claim that or would they incorrectly think, when overtaking, that increasing an interest in having the option to use a certain product in an illegal their speed enables them to avoid colliding with the oncoming vehi- way (without intending to do so) carries little moral weight when we cle? are deciding how to regulate the production and use of the relevant product. The case of assault rifles can be used for these purposes even For a splendid ‘all things considered defense’ of self-driving cars, if it is true that assault rifles and cars differ in that potential harm is even though they would not eradicate the existence of road accidents, a by-product of the latter but not of the former. We thank an anony- see [16]. Our proposal is much less radical, of course, than the pro- mous reviewer for pointing out the need for this clarification. posal to make self-driving cars mandatory. 1 3 AI and Ethics passengers would suffocate from misapplying them, e.g., ways that leave the option of harming people intact ([27], p. by getting the attached straps around their neck and the like. 215). Second, to the extent that mandatory implementation However, the fact that making safety vests mandatory on of LISA would involve a small cost in the form of limited ferries introduced this novel risk was (and is) not a good rea- erosion of the driver’s moral agency, this might be worth son on balance to decline to introduce them, since the risks it. Indeed, it would seem morally indecent to insist on the eliminated in this case—of drowning in an emergency—are avoidance of a small reduction of drivers’ moral agency at far graver than the novel risks introduced. Research since the cost of the deaths of thousands of people each week the 1990s has shown that the risks of associated with LISA ([27], p. 215). Similarly, it would be morally outrageous are very low, and that the rewards of requiring it to be fitted of the owners of assault rifles to resist, on grounds of the by law would clearly outweigh them ([27], p. 213). Fur- protection of their moral agency, mandatory fitting of Lim- thermore, car simulator experiments suggest that drivers in iting Targets Assistants on their assault rifles—e.g., in the cars fitted with LISA will tend to adjust their driving by presence of the infamous school shootings involving assault overtaking less (Jamson et al., p. 22). Supplementary and rifles with which we are all too familiar. Indeed, it seems mandatory programs educating drivers about when it is safe problematic to assign any moral weight at all to the preserva- to overtake in a car with the LISA system could be run (Jam- tion of an illegal option to engage in activities that are very son et al., p. 22). Finally, if LISA is combined with a system risky for other people. that “… would assist drivers in judging the safety of gaps in Brownsword’s argument also seems to show too much. the oncoming traffic stream, by means of a simple interface. Suppose cars could be redesigned so that they could not Such a support system could provide a valuable feedback be stopped by traffic barriers erected to protect pedestrians mechanism for drivers regarding the safety of overtaking (one sees these in cities now, around buildings vulnerable opportunities and, if combined with a mandatory ISA, could to terrorist attack). On Brownsword’s view, such redesign realize the traffic safety benefits reported elsewhere” ([12], would boost the moral agency of drivers, since it adds a mor- p. 22). ally unacceptable alternative to their set of options which, Second, Brownsword [3] worries that the fitting of LISA exercising moral judgment, they choose not to take. Clearly, to most vehicles would erode the moral agency and moral though, this would not justify redesigning cars in this way. responsibility of drivers. Indeed, he believes that what he There are good reasons why we want cars to continue to be calls “techno-regulation”, of which LISA is an example, in built in such a way that they can be stopped by traffic bar - designing out the option of agents inflicting harm on others, riers. The same is the case, we would argue, with building makes us passive moral agents “… that are no longer active cars so that they cannot travel faster than the speed limit. participants in a moral community” [3], p. 19). A third concern, also raised by Brownsword [3], is that In response to this, we note first that while LISA reduces the implementation of LISA for safety reasons is the first drivers’ moral agency in speeding, the moral agency of the step on a slippery slope that leads to a totalitarian society driver is still genuine. Drivers of LISA cars are, within the like the one described by George Orwell in his book 1984, legal speed limit, still in full control of overtaking, tail- in which the state uses technology extensively to regulate gating, and deciding to be or simply being inattentive in citizens’ behavior in detail. To our mind this worry about the potentially harmful effects of implementing mandatory LISA is speculative and, unlike the clear and immediate benefits of implementing Admittedly, none of this shows conclusively that LISA carries LISA, undocumented. It is important to remind ourselves no significant risks. However, in our view, in this context, as in the that the state already regulates products in all sorts of ways context of risk assessment of technologies, it is a positive, though not conclusive, reason to think that LISA does not introduce significant with an eye to preventing us from using them in ways that novel risks that studies to identify such risks have not yet identified cause harm to others. Typically, we are accustomed to these any. Should it turn out that overtaking with LISA is intolerably haz- regulated products. We embrace them and certainly do not ardous, a weaker form of LISA that would only prevent the car form worry that we are on a slippery slope leading to a totalitar- driving, say, 15% faster that the permitted speed, could reduce that hazard to tolerable levels. ian society. Few of us think that road barriers preventing Relative to the six Autonomy Levels (AL) for driverless vehicles, motorists from driving down a pedestrian street are a dan- drivers of LISA-enabled vehicles retain a high level of autonomy, as gerous concession potentially driving us in the direction of automated steering and acceleration is not part of LISA as we have illiberal societies like Nazi Germany, the USSR, etc. Finally, envisaged it. Actually, it is far from straightforward to place LISA it is worth keeping in mind that totalitarianism is first and on the United States Department of Transportation [28] six levels of autonomy. One could argue that LISA cars would sit at level AL 1, foremost about what the laws say (assuming there are laws) since those acting “… as the driver, are responsible for driving the and not about making citizens accept these laws. Indeed, vehicle. [However] when engaged, the system can perform either one could imagine raising the legal speeding limits, mak- steering OR acceleration/braking”. In a LISA vehicle, the system can ing installation of LISA mandatory, thus, as a consequence only be engaged in braking. 1 3 AI and Ethics reducing speeding and, thus, the number of traffic victims. LISA is enforced. On the latter issue, we note that, in many It strains credibility to think that this would put us on to a countries, by law cars need to be checked regularly by path toward a techno-totalitarian society. authorized mechanics, and that such checks could easily Fourth, mandatory fitting of LISA would make vehicles reveal if the LISA system has been adjusted. more expensive. Some drivers might even be unable to buy vehicles as a result. This, it might be argued, unjustly dis- advantages them relative to other drivers. Presently, fitting 5 Conclusion LISA hardware (including camera, brackets, trim and wir- ing) costs between €186–249 per vehicle. If LISA were We have argued that the speed of private vehicles—not to become mandatory, economies of scale would probably emergency vehicles (such as police cars, ambulances, and lower the costs of LISA. Moreover, the risks involved in fire engines) should be restricted by mandatory fitting of speeding are not equally distributed: disadvantaged people LISA. Our argument rests on a normative principle, the who cannot afford to buy a vehicle are subjected to them Principle of Required Prohibition, and on various empiri- without enjoying the benefits associated with its being per - cal premises regarding the effects of mandatory fitting of missible to impose this risk on others. Similarly, every fourth LISA and alternative means of reducing illegal speeding. traffic victim is a cyclist or a pedestrian. Yet they pose little We have argued that the Principle of Required Prohibition or no serious threat to others through their chosen means of is surprisingly weak in the sense of being very undemanding movement [27], 207–208). Hence, the appeal to fair distribu- and, thus, should be accepted by most if not all of us, and tion of benefits and burdens by those opposing mandatory we have appealed to various research reports to support the installation of LISA backfires: it also serves as an argument empirical claims we have made. Finally, we have rebutted against not addressing these unequal risks. Finally, manda- five objections to the legal prohibition of cars without LISA. tory installation would, as we pointed out above, result in Our conclusion is that the state should make it impossible a range of cost savings—e.g., on emergency services, hos- for drivers to illegally violate speed limitations by making pitals, insurance premiums. Surely, these savings could be the installation of LISA on cars compulsory by law, just as used to subsidize the roll out of LISA to the point that the the state, in many other ways, renders it impossible, or at present pragmatic concern is met. any rate harder, for us to violate laws. The case for manda- Fifth, and finally, some might contend that mandatory fit- tory LISA is, in all essential respects, just like the case for ting of LISA will not help much because drivers who want to road barriers that prevent motorists from driving on streets speed are likely to tinker with, or pay others to tinker with, reserved for pedestrians, or (to take a rather different issue) their speed limitation systems to enable them to drive faster the case for withholding taxes at source, thus depriving citi- than the speed limits permit—or make other adjustments if zens of the opportunity to under-declare their income and they are otherwise dissatisfied with how the system func- evade taxes they are required to pay by law. tions [4]. In support of this contention, they might point out Acknowledgements We would like to thank Jesper Ryberg, Sebastian how many (especially young) users of 50 cc mopeds have Jon Holmen, Sune Lægaard, Fatima Sabir, and two anonymous review- them altered so that they are capable of going faster than ers for very valuable comments. they were designed to go. Author contributions Both authors have contributed the same, con- If the mandatory fitting of LISA were to make no differ - cerning design, writing, review, etc. ence to speeding, the case for the scheme would fall apart. However, the antecedent here is very implausible. The great Funding Open access funding provided by Royal Danish Library. Funding was supported by Carlsbergfondet (CF21-0181) and The Dan- majority of 50 cc mopeds have not been tinkered with, and ish Research Foundation (DNFR-144). whereas “tuning” 50 cc mopeds involves a relatively simple mechanical intervention, tampering with LISA might well Availability of data and material Not applicable. be beyond the capabilities of all but a few trained mechan- ics with supplementary courses in electronics and software. Finally, how ineffective the LISA technology becomes is Admittedly, unscrupulous drivers could conceal their tamper- likely to depend on the fines for tampering with it, and more ing with LISA from the mechanics. The checking regime would be generally how stringently the prohibition on cars without imperfect in this respect. However, regular (annual?) adjustments turning LISA on and off are likely to be complicated and costly. How the state should implement obligatory installation of LISA on cars if that is made compulsory by law is an important topic which Based on individual costs extracted from calculations made in we have not dealt with in the paper. One model would require all 2012 (see https:// www. 20spl enty. org/ what_ is_ isa (accessed April 12, vehicles produced from, say, 2024 to have implemented LISA, as is 2023)). Admittedly, on top of the indicated costs comes the cost of the case with ASIA in UK and EU. Another model would require all the software for LISA. vehicles to have LISA implements at a certain date. 1 3 AI and Ethics 11. EU Regulation: /2144 https://eur -le x.eur opa.eu/ eli/ r eg/2019/ 2144/ Declarations oj. Accessed 21 Feb 2023 (2019) 12. Hegeman, G., Tapani, A., Hoogendoorn, S.: Overtaking assistant Conflict of interests There are no potential conflicts of interest to dis- assessment using traffic simulation. Transport. Res. Part C Emerg. close. Technol. 17(6), 617–630 (2009) 13. Hirst, D.: Regulating electric scooters (E-scooters). House of Ethics approval and consent to participate In the submitted article, Commons Library, London. Briefing Paper, 8958(10) (2021) there are no research involving human participants and/or animals and 14. Husak, D.: Vehicles and crashes: why is this moral issue over- therefore no informed consent from these non-existent research sub- looked? Soc. Theory Pract. 30(3), 351–370 (2004) ject—so not applicable. 15. Jamson, S., Chorlton, K., Carsten, O.: Could intelligent speed adaptation make overtaking unsafe? Accid. Anal. Prev. 48, 29–36 Consent for publication Both authors consent to the publication men- (2012) tioned above. 16. Kamm, F.M.: The use and abuse of the trolley problem: self- driving cars; medical treatments, and the distribution of harm. Open Access This article is licensed under a Creative Commons Attri- In: Liao, M.S. (ed.) Ethics of artificial intelligence, pp. 79–108. bution 4.0 International License, which permits use, sharing, adapta- Oxford University Press, Oxford (2020) tion, distribution and reproduction in any medium or format, as long 17. Lai, F., Carsten, O., Tate, F.: How much benefit does intelligent as you give appropriate credit to the original author(s) and the source, speed adaptation deliver: an analysis of its potential contribution provide a link to the Creative Commons licence, and indicate if changes to safety and environment. Accid. Anal. Prev. 48, 63–72 (2012) were made. The images or other third party material in this article are 18. Lee, Y.M., Chong, S.Y., Goonting, K., Sheppard, E.: The effect of included in the article's Creative Commons licence, unless indicated speed limit credibility on drivers’ speed choice. Transport. Res. F otherwise in a credit line to the material. If material is not included in Traffic Psychol. Behav. 45, 43–53 (2017) the article's Creative Commons licence and your intended use is not 19. Norcross, A.: Comparing harms: headaches and human lives. permitted by statutory regulation or exceeds the permitted use, you will Philos. Public Aff. 26(2), 135–167 (1997) need to obtain permission directly from the copyright holder. To view a 20. Oei, H.L., Polak, P.H.: Intelligent speed adaptation (ISA) and road copy of this licence, visit http://cr eativ ecommons. or g/licen ses/ b y/4.0/ . safety. IATSS Res. 26(2), 45–51 (2002) 21. Ori, M.: Why not road ethics? Theoria 86(3), 389–412 (2020) 22. Reed, N., Leiman, T., Palade, P., Martens, M., Kester, L.: Ethics of automated vehicles: breaking traffic rules for road safety. Ethics References Inf. Technol. 23(4), 777–789 (2021) 23. Regan, M.: Preventing traffic accidents by mobile phone users. 1. Agerholm, N., Tradisauskas, N., Harms, L., Lahrmann, H.: Pre- Med. J. Australia 185(11/12), 628 (2006) liminary results from the Danish ISA Project Spar på farten- 24. Sahebi, S., Nassiri, H., Van Wee, B., & Araghi, Y.: Incorporating Behavior (2007) car owner preferences for the introduction of economic incentives 2. Augeri, M.G., Cozzo, P., Greco, S.: Dominance-based rough for speed limit enforcement. Trans. Res. Part F: Traffic Psychol set approach: an application case study for setting speed limits Behav. 64, 509–521 (2019) for vehicles in speed controlled zones. Knowl. Based Syst. 89, 25. Shill, G. H.: Should law subsidize driving? NYUL Rev. 95, 498 288–300 (2015) (2020) 3. Brownsword, R.: Code, control, and choice: why East is East and 26. Shinar, D. (Ed.). Traffic safety and human behavior. Emerald West is West. Leg. Stud. 25(1), 1–21 (2005) Group Publishing (2017) 4. Brinkley, J., Dunbar, J., Smith, J.D., Gilbert, J.E.: A Usability 27. Smids, J.: The moral case for intelligent speed adaptation. J. Appl. evaluation of the BMW active cruise control system with “Stop Philos. 35(2), 205–221 (2018) and Go” function. In: Proceedings of the Human Factors and 28. United States Department of Transportation: Automated vehicle Ergonomics Society Annual Meeting, vol. 61(1), pp. 1536–1540. for safety. https:// www. nhtsa. gov/ techn ology- innov ation/ autom Sage CA: Los Angeles, CA: SAGE Publications (2017) ated- vehic les- safety. Accessed 21 Feb 2023 (2022) 5. Carsten, O.: Is intelligent speed adaptation ready for deployment? 29. Várhelyi, A., Hjälmdahl, M., Hydén, C., & Draskóczy, M.: Effects Accid. Anal. Prev. 48, 1–3 (2012) of an active accelerator pedal on driver behaviour and traffic safety 6. Elvik, R., Vadeby, A., Hels, T., van Schagen, I.: Updated estimates after long-term use in urban areas. Acc. Anal. Prevent. 36(5), of the relationship between speed and road safety at the aggregate 729–737 (2004) and individual levels. Accid. Anal. Prev. 123, 114–122 (2019) 30. WHO: Road traffic injuries. https://www .who. int/ ne ws-r oom/f act- 7. Elvik, R.: Speed limits, enforcement, and health consequences. sheets/ detail/ road- tra ffic- injur ies. Accessed 21 Feb 2023 (2022) Annu. Rev. Public Health 33, 225–238 (2012) 31. Yao, Y., Carsten, O., Hibberd, D.: An empirical approach to deter- 8. ETSC: Road safety target in sight: making up for lost time. https:// mining speed limit credibility. Transport. Res. F Traffic Psychol. e tsc. eu/ 4t h- annual- r oad- saf e ty- per f o r mance- index- pin- r epor t/. Behav. 63, 270–282 (2019) Accessed 21 Feb 2023 (2010) 9. ETSC: Briefing: intelligent speed assistance (ISA). https://e tsc.e u/ Publisher's Note Springer Nature remains neutral with regard to brie fing- intel ligent- speed- assis tance- isa/. Accessed 21 Feb 2023 jurisdictional claims in published maps and institutional affiliations. (2017) 10. ETSC: Reducing speeding in Europe https:// e tsc. eu/ r educ ing- speed ing- in- europe- pin- flash- 36/. Accessed 21 Feb 2023 (2019) 1 3

Journal

AI and Ethics – Springer Journals

Published: May 16, 2023

Keywords: AI and driving; Ethics of legal regulation; Limited intelligent speed assistant (LISA); Speed limits; Risk imposition; Traffic

There are no references for this article.