“[…] who will maintain our families, whilst we undertake the arduous task [of learning some other business]; and when we have learned it, how do we know we shall be any better for all our pains; for by the time we have served our second apprenticeship, another machine may arise, which may take away that business also; so that our families, being half pined whilst we are learning how to provide them with bread, will be wholly so during the period of our third apprenticeship.”
Leeds Woollen Workers Petition, 1786.
Abstract
Debates concerning digital automation are mostly focused on the question of the availability of jobs in the short and long term. To counteract the possible negative effects of automation, it is often suggested that those at risk of technological unemployment should have access to retraining and reskilling opportunities. What is often missing in these debates are implications that all of this may have for individual autonomy understood as the ability to make and develop long-term plans. In this paper, I argue that if digital automation becomes rapid, it will significantly undermine the legitimate expectation of stability and consequently, the ability to make and pursue long-term plans in the sphere of work. I focus on what is often taken to be one of the main long-term plans, i.e. the choice of profession, and I argue that this choice may be undermined by the pressure to continuously acquire new skills while at the same time facing a diminishing range of professions that one can choose from. Given that the choice of profession is significant for not-work related spheres of life, its undermining can greatly affect individual autonomy in these other spheres too. I argue that such undermining of individual planning agency constitutes a distinctive form of harm that necessitates a proactive institutional response.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
Digital technologies have penetrated many spheres of our lives and will continue to do so, perhaps to an even greater extent, in the future. To name a few, most of us with access to the Internet interact with algorithms on a daily basis when we read the news, order products or services, or search for potential romantic partners. We are also getting used to the presence of automated technology in our physical environment be it vacuum robots in our houses, semi-automated cars on the streets or drones in the sky. Furthermore, and similarly to the signatories of the 1786 Leeds Woollen Workers Petition (better known as Luddites) quoted above, we are becoming more aware of predictions concerning the automation of an increasing number of jobs, making us wonder when our own job will make it to that list, given that there is a fair chance that it will sooner or later (Frey and Osborne 2017; Susskind 2022). Technological developments such as these are indicators of what some call the Fourth Industrial Revolution.Footnote 1
It is hard to deny that the unfolding technological innovations are revolutionary given their velocity, scope and complexity. As opposed to the previous industrial revolutions, the fourth one is evolving not at a linear, but at an exponential pace and is disrupting almost every industry in every country in ways that are often difficult to comprehend and anticipate (Schwab 2017).Footnote 2 To be sure, the previous revolutions also caused major disruptions at the time, but those disruptions were temporally limited and societies could go back to some form of equilibrium after a transitional period.Footnote 3 What makes the Fourth Industrial Revolution distinctive is that due to the exponential pace of its evolvement, it may lead to perpetual social disruption.Footnote 4 According to some estimations, the level of disruption may be ten times faster and 300 times the scale of the impact of the First Industrial Revolution (Dobbs et al. 2015).Footnote 5 Therefore, there is a great likelihood that the Fourth Industrial Revolution will, in the words of Klaus Schwab, Founder and Executive Chairman of the World Economic Forum (2017: 6), “fundamentally alter the way we live, work and relate to one another”.
The debate about how these developments impact human autonomy has been growing exponentially too. Much of it is focused on what can be called a micro-level—examining the direct effects of particular technologies on specific aspects of human autonomy. It is argued that algorithmic decision support systems and predictive analytics can and often do interfere with individual decision-making via, for instance, personalizing the individual choice environment or manipulating the way people make choices. They do so by collecting and analyzing large amounts of personal data to conduct targeted interventions aimed at influencing individual behaviour (Benn and Lazar 2022; Danaher 2019; Müller 2021).Footnote 6 Thus, algorithms not only affect the range of options people can choose from, but they can even affect the way those choices are made—in a less deliberative and more arbitrary way, which in the long term may decrease the overall capacity for decision-making (Laitinen and Sahlgren 2021). While all these considerations merit the increasing level of attention they receive, there are developments at the macro-level too that have received far less attention so far.Footnote 7 By macro-level, I mean the broader social and political effects of technological development taken as a whole at individual and collective levels.Footnote 8 Technological innovations are sometimes characterized as socially disruptive since they not only disrupt particular social interactions, but they are also increasingly disrupting the functioning of societies by and large.Footnote 9 While the implications for collective autonomy surely deserve attention on their own, here I will focus on the impacts of social disruption on individual autonomy.Footnote 10 If individual autonomy is about developing and pursuing long-term plans, how is this affected by the constant disruption that the increasing deployment of digital technologies may face us with? More fundamentally still, what does it mean to live an autonomous life in the face of the perpetual disruption, and instability and uncertainty that may come with it?
I will address these questions by focusing on one of the major drivers of the Fourth Industrial Revolution—digital automation. As is well known by now, employers are increasingly introducing various forms of computerization in order to decrease the costs of production by either increasing the efficiency of human labour or by replacing it altogether (Müller 2021). Digital automation has generated a lot of debate concerning the possibility of technological unemployment, that is, whether computerization will lead to an overall decrease or increase in jobs available to humans (Autor 2015; Susskind 2022). To counteract possible negative effects of automation, it is often suggested that those at risk of technological unemployment should have access to retraining and reskilling opportunities (Acemoglu and Autor 2012; Autor 2015; Goldin and Katz 2007; Goos 2018; Willcocks 2020). However, what is missing in these debates is to consider how all of this affects individual planning agency. Although we do not yet witness large-scale labour transformations, given the pace of technological development this may soon be the case.Footnote 11 If digital automation gets deployed at a rapidly accelerating pace, how will this affect the formation and pursuit of long-term plans in the sphere of work and beyond?Footnote 12
In the paper, I argue that if digital automation reaches this point, a whole new set of problematic issues may arise. More specifically, I argue that rapid digital automation may undermine what I define as the legitimate expectation of stability, and consequently, individual ability to develop and pursue long-term plans in the sphere of work. I show this by focusing on what is often taken to be one of the main long-term plans people may have—the choice of profession. I argue that this choice may be undermined by the pressure to continuously acquire new skills while at the same time facing a diminishing range of professions that one can choose from. Given that the choice of profession is significant for not-work related spheres of life (Rawls 1971; Raz 1988), its undermining can greatly affect individual autonomy in these other spheres too. Overall, undermining of individual planning agency constitutes a distinctive form of harm that necessitates a proactive institutional response. Therefore, the impact of automation on the labour market should not be assessed only in terms of whether the number of jobs is increasing or decreasing, but we also need to think about what those jobs will be like, what people will be required to do in order to get or preserve them, and how fast people may be forced to change them.
2 Autonomy and Expectations
It is not an overstatement to say that individual autonomy is a core liberal value. While autonomy can be defined in various ways, here I understand it as the ability to develop, revise and pursue one’s plans and commitments. The ability presupposes appropriate mental faculties; an adequate range of options to choose from and freedom from coercive or manipulative interference by others with our options and choices.Footnote 13 To live an autonomous life, it does not suffice that a person has the ability, but they need to actually exercise it to decide about the life they want to live (Raz 1988: 372–73). While an autonomous life includes both short-term and long-term plans, here I predominantly focus on the latter as pursuing long-term plans is often associated with leading a good life (Goodin 1995; Rawls 1971).Footnote 14 Importantly, within a liberal framework, a good life is not characterized by pursuing any particular long-term plan. Rather, the validity of the long-term plans lies in their authenticity—in the fact that both the plan and the means to attain it are at least partly chosen by the agent herself (Raz 1988).Footnote 15 To illustrate the plausibility of this point, think about the way many of us felt during the Covid-19 pandemic. Even if they set goals, it was very difficult for many people (especially those with small children) to work towards them due to the constant stream of interruptions (Szende 2020). It is hard to deny that the failure to pursue one’s plans or to make them significantly affects how well one’s life goes.
We develop our long-term plans based on certain expectations we have about the future state of affairs. Expectations are predictions that help people identify options open to them (Meyer and Sanklecha 2011). While we typically have a range of expectations, here I focus on those that have been characterized as legitimate. To be legitimate, expectations need to be epistemically valid in the sense that people holding them need to have good reasons to believe that the expectations will be fulfilled.Footnote 16 Legitimate expectations are also normatively relevant in the sense that they generate a justifiable claim for some form of response. For instance, imagine three friends playing a three-player board game every Wednesday evening. One Wednesday evening, one of them does not show up without any explanation. It is plausible to say that the other two friends had an epistemically valid expectation that the third one will show up, and that such an expectation is normatively relevant since it significantly affected their plans for that evening (they could not play the game without the third person). Thus, they are owed a response from the one who failed to show up, be that an explanation or apology (or both).Footnote 17 Finally, legitimate expectations are those that are induced by institutions. Since this condition is what ultimately differentiates legitimate expectations from other forms of expectations, let me spend some words explaining it.Footnote 18
Rawls (1971) famously argued that the rules of the basic structure determine the legitimacy of expectations; expectations formed under an unjust basic structure cannot be legitimate and consequently, their frustration is not normatively significant. Soon after A Theory of Justice has been published, Buchanan raised an important objection that just institutions are inherently unstable since the difference principle demands continuous institutional revisions in the light of technological developments in order to find better ways to improve the position of the worst off. Such institutional changes cause “gross and frequent” disruptions of an individual in framing and executing long-term plans as they undermine the individuals’ expectations, disrupt their lives, and ultimately, interfere with their rational agency (Buchanan 1975: 420).Footnote 19 Therefore, just institutions are not all that matter; rather, the stability of expectations formed under those institutions matters too. This point has been further developed in more recent debates about the legitimacy of expectations induced by non-ideal institutions.
It is argued that at least some expectations can be legitimate even under non-ideal conditions since they are induced by institutions making promises, giving assurances or by the continuous provision of a benefit (Brown 2011). According to Brown (2012: 634), “a legitimate expectation is not the mere subjective expectation of a benefit […] rather, it is a claim to the benefit based on an objective indicator that has been or can be publicly recognized.” In addition, expectations induced by institutions play an important role in our lives since at least some of them are crucial for making and pursuing long-term plans and commitments. The frustration of such expectations is not merely about not satisfying a certain desire but has adverse effects on one’s functioning as “a rational planner and executor of plans” (Buchanan 1975: 420). It follows that the harm caused by frustrating legitimate expectations is a distinct type of harm since the person is made worse off in a distinct sense: their successful pursuit of long-term plans based on those expectations is undermined and they are left worse off than they might have been had they not formed such expectations (Brown 2012; Meyer and Sanklecha 2014: 371).Footnote 20 Here we arrive at the key point: public institutions commit injustice when they induce and then frustrate legitimate expectations.Footnote 21 The injustice constitutes what Simmons (2010: 20) calls “rug-pulling”: institutions unfairly pull the rug from beneath those whose expectations that particular rules will remain unchanged they have induced (also Feinberg 1973). Since rug-pulling is unfair, those whose legitimate expectations are frustrated, are owed compensation. The point of compensation is not to slow down reforms, but to fairly distribute their burden and as such, it is compatible with evolutionary institutional changes (Meyer and Sanklecha 2014).Footnote 22 The institutional liability for damages has an important stabilizing function—the knowledge that institutions are liable for frustrating legitimate expectations makes those expectations more stable, thus enabling individuals to pursue medium- to long-term plans (Brown 2011; Goodin 1995).Footnote 23
While the reviewed arguments make a strong case that the frustration of legitimate expectations constitutes a distinctive form of harm due to the consequences it has for individual planning agency, it remains unclear what exactly individuals can reasonably expect and what are appropriate institutional responses in case such expectations get frustrated. As Räikkä (2014) points out, it is inherent to institutional rules and practices that they change in order to respond to changing circumstances.Footnote 24 In his view, it may not be reasonable to expect that particular rules remain unchanged; but it is reasonable to expect that the rules and practices do not change frequently and abruptly since that kind of change undercuts individual planning abilities. It follows that not any kind of rug-pulling is unfair; instead, we should be concerned with a “constant rug-pulling” (Räikkä 2014: 26). The notion of “constant rug pulling” captures something important for the present discussion: as opposed to one-time rug pulling that may affect particular expectations and related particular long-term plans, constant rug pulling can cause another, more serious form of harm that concerns the undermining of the very ability to make plans. For it is one thing to be forced to replace one long-term plan with another one, and it is another not to be able to develop any long-term plan whatsoever.Footnote 25 To capture such a form of harm, I will refer to the expectation frustration of which leads to such harm as the expectation of stability. The expectation of stability is ultimately about expecting institutions to provide some degree of stability in the lives of those subjected to them (Goodin 1995; Räikkä 2014).Footnote 26 This means that when conditions significantly change, it is legitimate to expect institutions to adjust the rules or even adopt new ones to lessen the disruptive effects of the changes. So defined, the expectation of stability is closely related to the principle of the rule of law as the central justification of the rule of law is that it provides a stable set of rules based on which agents can form expectations about the future, and plan and execute their plans accordingly (Green 2017).Footnote 27
While the expectation of stability is significant for various spheres of life, here I will focus on employment. In the sphere of employment, the expectation of stability is typically related to professions as one of the main long-term plans people may have, at least in societies in which paid work is the main way for many to make a living. I will show how rapid digital automation can undermine the notion of professions and consequently, people’s ability to make long-term plans in this respect.
3 Automation and the Expectation of Stability
Automation has been said to generate two possible negative effects on the labour market. The first effect is the risk of technological unemployment: the concern that technological innovations may lead to continuously decreasing the number of available jobs.Footnote 28 The issue is not distinctive of the Fourth Industrial Revolution but was also raised concerning previous industrial revolutions.Footnote 29 While there is consensus that the deployment of technological innovations can lead to a temporary decrease in the availability of jobs, the jury is still out on determining its long-term effects on the labour market (Autor 2015; Schwab 2017; Willcocks 2020).Footnote 30 However, even if the aggregate number of jobs would increase, it is undeniable that the meaning and the value of work as well as its role in society and our individual lives will be significantly transformed (Santoni de Sio et al. 2021; Willcocks 2020). This takes us to a second issue that seems to be distinctive of the Fourth Industrial Revolution.
According to some economists, the problem is not that the broad deployment of technological innovations brings a risk of the decreasing availability of employment in the long run (although this may be the case); rather, the problem is that automation affects the quality of jobs for a significant proportion of workers (Autor 2015). This trend has been described as the polarization of jobs into two extremes—those requiring high skills (non-routine cognitive and abstract tasks) and those requiring low skills (non-routine interactive and manual tasks) (Goos 2018). As computerization progresses, it has been observed that the demand for jobs at these two extremes is growing, but jobs that used to be in the middle are disappearing.Footnote 31 The hypothesis of routine-biased technological change (RBTC) stipulates that routine-based labour tasks, typically clustered in middle-skill jobs, such as administrative support and sales, are the easiest to automate as humans can explicate what these tasks involve and codify it in software (Acemoglu and Autor 2012).Footnote 32 Job polarization is followed by wage polarization too since the deployment of technological innovations increases demand for highly skilled labour to effectively use them, but such labour is insufficiently supplied since educational systems do not keep up with technological innovations and it also takes time and effort to acquire those skills (Autor 2015; Autor and Dorn 2013; Goos 2018).Footnote 33 To counteract this trend, it is suggested that significant educational reforms need to take place in order to secure retraining and upskilling workers so that the labour supply can keep up with the demand (Acemoglu and Autor 2012; Autor 2015; Goldin and Katz 2007; Goos 2018; Willcocks 2020). While the concern to address the increasing income inequality is surely justifiable, it is also important to examine how the race between education and technology may affect the expectation of stability that is necessary for individuals to develop and pursue their long-term plans. If the claim that automation may take a more accelerated pace is correct, then we should not only be concerned about the number of available jobs but also about the impact of the fast-changing working (and educational) environment on the individual ability to develop and pursue long-term plans.
When it comes to the sphere of work, the most significant long-term plan a person can make concerns the choice of profession.Footnote 34 The significance of the choice of profession goes beyond the sphere of work given how much it affects other aspects of one’s life too.Footnote 35 To say that the choice of profession is a long-term plan does not mean that people never change their professions. As mentioned before, long-term plans are not fixed in any sense, but people revise their plans or the means to pursue them in the light of changing circumstances (Rawls 1971). Even when we make a radical break in our life, however, such as deciding to enter another profession, we expect that all else equal, we can stay in the new profession. The expectation that one can have a profession is induced by the existing institutions through numerous regulations including those in the sphere of education and labour, based on which people make choices of professions.Footnote 36 If anything, the very definition of what counts as a profession and the sets of skills it involves is often officially regulated through, for instance, the accreditation of degree programs or by issuing licenses to practice particular professions. Moreover, the freedom of occupation is guaranteed by many constitutions worldwide. Therefore, the very idea that there is such a thing as a profession as well as the idea that one can pursue it if one wants and can, is induced by state institutions. However, if the deployment of digital technologies becomes as rapid as assumed here, it may undermine the notion of professions in two major ways: indirectly, by continuously restructuring skills that constitute professions, and also more directly, by diminishing the range of professions people can choose from.
3.1 Fragmentation of Professions
Technological change affects the nature of tasks available to humans, and consequently, the nature and range of skills necessary to perform those tasks (Autor 2015). This is part of the ongoing process that Susskind (2022) defines as “task encroachment”; that is, technology gradually but “relentlessly” taking up more and more tasks. The more tasks are automated, so it is argued, the more human labour will need to be upskilled in order to keep up. For instance, Willcocks (2020) estimates that in a 10-year time horizon, every person’s job is likely to be changed by at least 25%, which entails acquiring a new set of skills to fill in a newly emerged gap. Note that these predictions refer to the short term; in the longer term, the changes will likely be much greater. The increasing need for upskilling can affect the ability to make long-term plans in the sphere of work in several ways.
Let’s start from the very notion of upskilling as such. If the pace of digital automation becomes rapid, it is likely that to keep up with it, one retraining may not suffice; rather, people may be expected to enter a lifelong cycle of retrainings.Footnote 37 To illustrate, Willcocks (2020: 287) says that individuals need to adopt a “continuous learning ethos and think through the skills that are going to be in demand at various points in their work careers.” To be sure, there is nothing inherently problematic with the need to update one’s set of skills or acquire new ones. On the contrary, it is part and parcel of living an autonomous life to be able to revise one’s long-term plans should circumstances change. However, such revisions come at certain costs.
Recall that one of the main Luddites’ worries was that they may end up in a vicious circle of perpetual training whereby skills they develop may become quickly devalued due to the invention of new machines, thus forcing them to acquire new skills that may also be susceptible to devaluation.Footnote 38 Luddites had correctly pointed out that one of the obvious downsides of being caught in the cycle of retrainings is that these necessitate resources at both individual and collective levels. Unless certain measures are put in place, people may need significant resources in order to afford cycles of retrainings.Footnote 39 To be sure, how much retraining one may need depends on their initial set of skills. Consider a former electrician who needs to retrain to become a solar panel installer. The difference in the existing and needed sets of skills is much smaller as compared to someone who was trained as an accountant but had to become a lorry driver, only to end up as a retail assistant.Footnote 40 From the perspective of long-term planning, the latter is more objectionable than the former. But the former may be objectionable too if it turns out that the person may need to undergo more than one retraining since the difference between the initial set of skills and the skills acquired through additional training may keep increasing. After undergoing a certain number of changes, it is questionable whether we can still talk about the same long-term plan, or if the person is pursuing a completely new one.
Moreover, to make revisions concerning a set of skills a person has, one needs to be able to anticipate what kind of retraining is needed; but in the conditions of accelerated change such predictions are very hard to make. This is not limited to reskilling but soon it may concern making decisions about what kind of education to pursue as well. One of the preconditions for making autonomous choices is that to choose, a person must be aware of their options (Raz 1988: 371). If we cannot know what options there are or will be, then obviously it is very difficult to make any choices, let alone more informed ones, about reskilling or pursuing a particular educational path. At this point, one may object that we should not be focusing on the level of individuals; rather, it is the task of institutions, including educational ones, to anticipate what kind of retrainings and reskilling will be needed and provide those. Indeed, this is one of the key claims that Goldin and Katz (2007) make: that the educational system needs to provide education such that people can acquire general skills that are transferable across industries, thus making them less vulnerable to the changes in the labour market caused by technological changes.Footnote 41 While this is a valid point to make, it assumes that educational systems can race with technology, but this is conditional upon the speed of the race; the faster the technology race, the more difficult it will be for educational systems to keep up.Footnote 42 Moreover, Goldin and Katz have shown that educational systemsFootnote 43 were lagging behind technology already in the 1990s which makes one wonder how they can catch up with the newest developments in AI and robotics?Footnote 44 This is not to say that educational reforms should not take place; the point I want to make is that the aim of reforms should be to strengthen people’s ability to make and pursue long-term plans. As such, the point can provide additional normative support for the claim that educational reforms need to be such that they secure access to general transferable skills for everyone affected. I will come back to this.
I mentioned earlier that those who decide to enter and pursue a particular profession also expect to stay in that profession for some time at least. Although there are various reasons why this may be the case, here I focus on a particular reason that illuminates an additional type of cost that the continuous change of tasks and skills generates. This has something to do with what Rawls (1971: 426) called “the Aristotelian principle”: the more proficient we are in something, the more we want to do it, and the more we enjoy doing it. Striving for excellence typically requires sustained effort and focus. By drawing on a range of empirical work, Gheaus and Herzog show that people are often motivated to work in order to achieve some level of excellence, and this includes not only the development of a particular set of skills but also the accomplishment that results from the exercise of these skills, which they consider as one of the important non-monetary goods of work. To achieve a certain level of excellence, employees need to be given enough time to develop solid skills and use them (Gheaus and Herzog 2016). Continuous disruptions and a need to learn new skills may make such enjoyment in perfecting and using particular skills impossible to obtain. Things may be even worse for those who, despite possessing one set of complex skills, may not be able to acquire another set of complex skills should the initial skills become devalued.Footnote 45 For instance, studies show that those with middle skills upon losing their jobs are more likely to compete for the low-skill ones due to the high costs of acquiring high skills (Autor 2015; Goos 2018). Those who find themselves in such a position face an additional cost of not only not being able to enjoy doing the complex task that one used to do, but of not being able to do any complex tasks anymore given an inability to acquire a new set of complex skills. Falling back from doing a complex task to doing something simpler arguably reduces one’s well-being.Footnote 46
As we can see, the idea that retrainings and reskilling may be an appropriate response to rapid digital automation overlooks the various costs that these entail. This brings me to a central problem that the approach may face—namely, the costs of retrainings may be unfairly distributed across the workforce. Economists point out that one of the main effects of technological change on the labour market is that it transforms jobs in a way that human labour complements work done by machines (Autor 2015). The problem arises when workers who are displaced by machines cannot take up complementary tasks due to significant differences in skills (Susskind 2022). Thus, the complementarity effect although often pictured in a positive light has a disproportionately harmful impact on people who either do not have such skills or it would require a lot of monetary and non-monetary resources for them to acquire those skills. Consider further that the burdens of retraining may not be fairly distributed across generations either.Footnote 47 As Danaher notes, the accelerating change may push members of younger generations into the “cycle of immiseration”: not only that they are more vulnerable to technological unemployment, but they may also lack resources to acquire higher skills since part-time jobs that often enable many of them to pursue higher education, may be automated too. Hence, besides the increased susceptibility to technological unemployment, younger generations may also have a decreased access to reskilling (Danaher 2019: 40–42).
3.2 Diminishing Range of Professions
As we can see, automation can affect professions in terms of restructuring and fragmenting the sets of skills they presuppose. But there are other, more significant ways in which professions can also be affected. Probably the most significant one is that the range of professions to choose from may become largely diminished. While it is certainly difficult to predict which new professions may emerge due to technological innovations, it is safe to assume that many of them will necessitate at least some degree of technical skills. The reason for this is that, as mentioned earlier, automation proceeds to a significant extent by introducing machines to complement human labour by taking up particular tasks (Autor 2015).Footnote 48 What this essentially means is that in the best-case scenario, humans will be working alongside intelligent machines. To be able to do so, they need to have certain technical skills, although the level of proficiency will vary across different types of jobs. Therefore, even if we assume that new professions will emerge in the long term, it is not an overstretch to claim that many of them will be technology oriented. It is not clear at all that the range of similar options, irrespective of how broad it may be, constitutes a context for a genuine, autonomous choice. I mentioned earlier that one of the preconditions for living an autonomous life is to have an adequate range of options to choose from. While it is debatable what makes a range of options adequate, it is difficult to deny that adequacy at least necessitates the variety of options. Those who would like to buy a house do not have a choice if they need to choose between a hundred identical houses (Raz 1988: 375–380). If more and more professions become digital technology-oriented, the variety of options to choose from will diminish accordingly. An additional point of having an adequate range of options is that a person can deliberate about them and by doing so, exercise their agency by choosing one option and rejecting other available options (Hurka 1987). However, if the variety of options keeps diminishing, then there is less and less room for people to exercise their judgment by choosing between them up until the point where there would be no space to make any judgment at all. Add to this the fact that the range of available options is increasingly determined by those who design technological innovations and make significant profits from their deployment. For instance, those who have designed recently released image or text-generating AI have significantly affected not only the future of the photography and writing-related professions but also the present of those who are in these professions right now. As it is not clear how those affected can counter such developments, it is difficult to resist the thought that their range of options has been interfered with by much more powerful actors able to bring about such changes.Footnote 49
Note a further complication. If the empirical claim concerning the rapid pace of digital automation and related disruptions is correct, it becomes questionable to what extent professions (as inherently long-term projects that necessitate the longer period of time to acquire and exercise the necessary skills) as such will survive. Since professions are the ground for making long-term plans in the sphere of work, individuals may no longer be able to form such plans. Instead, choices related to work may be narrowed down to a series of ad-hoc decisions to take up tasks that machines cannot yet do. As opposed to the burden of retrainings that I argued earlier is disproportionately carried by low-skill workers and younger generations who have fewer opportunities for meaningful reskilling, all those for whom their labour is (or will be) the main source of making a living are affected by the diminishing range of professions to choose from.Footnote 50
Therefore, wide deployment of digital automation may significantly undermine the professions by not allowing enough time to obtain and exercise one’s profession or by limiting the range of professions. As long as work remains the main source of making a living for many, it seems that people face a risk of being deprived of many of the valuable things the more stable professions bring with them, including the opportunity to make long-term plans as well as personal satisfaction from developing and exercising certain skills. To make it worse, some people may be deprived of more things than others through no fault of their own. Therefore, by fragmenting professions and decreasing their range, those who deploy rapid digital automation threaten all three aspects of individual autonomy in the sphere of work: limiting the range of options to choose from, interfering with freedom of choice as well as undermining the very ability to make and pursue long-term plans. Thus, the impact on human autonomy may be greater than it is acknowledged.Footnote 51 Note that one need not agree that a good life must include some degree of planning in order to accept the point I make here. Earlier I mentioned that the importance of planning agency goes beyond normative accounts of autonomy; planning agency is important to understand human action. Given that planning norms, as shown by Bratman (2018), are constitutive of practical rationality, one may worry that rapid changes that undermine planning agency, are also undermining human practical rationality to a significant extent.
I argued earlier that frustration of legitimate expectations constitutes a distinctive form of harm that necessitates institutional remedy. Before considering what institutional remedies are appropriate in this case, I first need to address two pressing objections.
One may object that the view defended here is controversial as it presupposes the value of work in individual lives, but not everyone would agree about this. Some people may value their free time for various reasons and may prefer living in a world in which they do not have to work in the traditional sense of the word, but can pursue their hobbies unconstrained by the need to make a living out of them. Thus, the claim concerning work as the source of a meaningful life amounts to endorsing particular conceptions of the good life and as such may not be compatible with liberal accounts of autonomy that I take for granted here (Arneson 1987; Parr 2022). I do think this is a powerful objection to raise as indeed not everyone would agree that work is one of the main sources of meaning in life. However, the claim about the significance of professions for long-term planning does not presuppose any claim concerning the value of work. Even in a post-work society, people would presumably do something even for leisure, and what matters for my argument is whether they can do it continuously so that it can constitute a long-term plan. Therefore, my argument is not about having stable access to professions and employment as such, but it is about having stable access to things that constitute the basis for our long-term plans.
The second objection is that the expectation of stability concerning the sphere of employment is not epistemically valid as people should know that such transformational changes are underway and adjust their plans accordingly. In support, one could point to the increasing degree of flexibilization of the labour market as well as general awareness about ongoing technological development. Therefore, it may appear that instead of stability, it is reasonable to expect instability.Footnote 52 In response, I argued earlier that expectations (including legitimate ones) are necessary for people to make and pursue their plans. To say that people should start expecting a more unstable environment is to say that they should give up on their planning agency. But as Bratman (2018) shows it is not clear that we can do so given that planning is an important component of human practical rationality. That said, the objection does make an important point that socially disruptive technologies do not disrupt individuals only but also disrupt institutions. How can we expect increasingly disrupted institutions to secure some level of stability in the lives of those subjected to them? I turn to this now.
4 What Should Institutions Do?
I mentioned earlier that the expectation of stability is inherently related to the principle of the rule of law. The rule of law is not limited to mandating the continuation of the existing regulations, but also entails adopting new ones in order to provide some level of stability in the lives of those subjected to those institutions. After all, the value of stability is one of the moral goals that public institutions need to secure. To be sure, it is not the only one and can be overridden by other morally worthy goals, but it is at least a pro-tanto goal that state institutions should pursue (Goodin 1995). We could also see that the frustration of legitimate expectations amounts to a distinctive form of harm and as such, entitles those harmed to some form of compensation, all else being equal. The compensation is aimed at securing individuals’ expectations while at the same time enabling institutional reforms to take place. Again, the purpose of compensation is not to prevent institutional reforms but to distribute their burden more fairly (Brown 2011; Trebilcock 2015). In this sense, compensating for the harm induced by changes can be characterized as a reactive approach—reacting to changes that are already taking place.
Applied to the case of digital automation, the reactive approach would encompass various policy proposals. I mentioned earlier that many economists think that for human workers to keep up with fast-developing machines, they need to get appropriate training. This further entails that educational institutions would need to be reformed to provide such trainings. What the argument developed here entails for such proposals is that they need to be compatible with the individual ability to make and pursue long-term plans. For instance, this would favour trainings that secure a more robust set of skills over those that may expose people to several cycles of retrainings. However, I also mentioned one important limitation of this approach—that it presupposes that education can race with technological developments, but this assumption may be unwarranted. There is a more general issue here that has something to do with the reactive approach as such—namely that attempts to respond to ongoing technological changes by trying to adjust may be increasingly inadequate.Footnote 53 Namely, compensation is justified only if it can ease the effects of changes on the affected individuals. Going back to the expectation of stability, it would mean that an individual’s expectations can be stabilized to the degree that they can make and pursue long-term plans. But what if changes are so significant that it is unclear what kind of compensation is adequate for the harm they induce? Should this be the case, then instead of providing inadequate compensation, institutions need to take a more proactive approach and attempt to decrease the risk of the expectation of stability being frustrated.Footnote 54
The proactive approach would favour a different (although compatible) set of policy proposals, such as introducing automation taxes in order to counter the earlier mentioned disproportionate effects that rapid automation can have on some people. The proposal has been criticized on the grounds of slowing down economic growth and decreasing potential benefits that may come from it (Parr 2022), but it is not clear why these should be taken for granted.Footnote 55 If anything, in the context of climate change, there is an increasing number of voices arguing in favour of curtailing economic growth (and consumption) and going in the opposite direction of degrowth (Kallis 2018).Footnote 56 Whether or not we agree with such proposals, their proponents are right in one thing—the value of economic growth should not be taken for granted. Moreover, the abovementioned reactive strategies that include more robust educational reforms or the introduction of UBI necessitate a lot of resources and these resources need to come from somewhere. If human labour keeps decreasing, then revenues from income taxes will decrease too and will need to be supplemented by revenues from other sources. This is not to argue in favour of these more radical proposals since much more needs to be said in their defence than I can say here, but only to point out that the current trend in design and deployment of technological innovations rests on assumptions and value judgments that should not be taken for granted. This brings me to a main point I want to make here: the social and political implications of digital technologies are too large to be isolated from public debate and democratic decision-making. Digital technology generates new power dynamics that raise new demands for procedural legitimacy (Lazar 2022). What this essentially means is that the impact of digital technologies should not be the only subject of institutional regulation; instead, their design as well as decisions concerning their deployment also need to be subjected to procedural legitimation. Going back to the impact of digital automation on professions, one question that should be open to public debate is which professions to automate since, in all likelihood, many (perhaps most) of them can be automated sooner or later.
Even if one can make a case that public institutions need to take a more proactive approach and decrease the risk of frustrating the expectation of stability, what if it is no longer feasible (Danaher 2019)? Given the decreasing regulatory power of institutions due to the disruptive nature of the technology as well as the increasing power of non-state actors, one may wonder what role is left for public institutions in the future (Schwab 2017).Footnote 57 The general trend seems to be not toward more stable institutions able to navigate changes, but toward more flexible people adjusted to a less stable institutional background. However, if they are to be justified to those who are subjected to them, public institutions need to address a growing number of issues raised by digital technologies. Therefore, the ability (and willingness) to regulate the sphere of digital technologies is becoming crucial for the legitimation of public institutions under contemporary conditions. This shows that the deployment of socially disruptive technologies challenges the very foundations of liberal political thought.Footnote 58 Thus, all those who value individual autonomy and think that institutions should play an essential role in securing it, need to wrestle with this problem. Moreover, we need to wrestle with this problem while we still can.
Notes
The revolution is far more encompassing, including 3D printing, nanotechnology, biotechnology, energy storage, quantum computing, etc. I do not consider these here.
The three previous industrial revolutions concerned the use of water and steam to mechanize production; the use of electric power to create mass production; the use of electronics and information technology to automate production; and the fourth is about the use of AI and machine learning to further increase production (Schwab 2017).
A paradigmatic example is the First Industrial Revolution which brought about tectonic changes by moving labor from rural areas and agriculture to urban areas and manufacturing, which led to the creation of the working class (Hedman 1989).
For instance, the computer processing power, which is behind AI, doubles every 2 years (Colvin 2015).
A word of caution is needed here. The social (and political) disruption is related to the massive deployment of digital technologies. However, there is a long road before innovations get diffused and broadly deployed in practice (Willcocks 2020). This means that the examination of potential social disruptions resulting from the broad deployment of digital technologies rests on inherently speculative grounds.
The distinction between micro- and macro-levels broadly maps onto the distinction between the interactional and structural approach to automated influence as developed by Benn and Lazar (2022). They plausibly point out that the dominant focus on the interactional dimension of automated influence misses its structural dimension.
The impacts of various digital technologies are typically assessed in isolation from each other; however, it is important to assess their cumulative effects too.
Probably one of the most salient social disruptions concerns the change of power relations. As Lazar (2022) points out, digital technologies are intricately connected to power as they are used to exercise power, they have enabled new power relations and also intensified the existing ones thus triggering legitimating demands.
In the debates about the impact of automation on the labor market, there is a different sense of what counts as rapid change. For instance, Willcocks (2020) argues that the widespread diffusion of innovation takes up four to five years, and he characterizes this as “slow” change. However, changes that may take four or five years are not necessarily slow, especially when it comes to collective levels.
This kind of inquiry is inherently speculative which means that the argument defended here is conditional upon the empirical claims concerning the pace of automation being true.
The significance of the ability to develop and pursue plans goes beyond the normative accounts of personal autonomy. As Bratman (2018) shows, human agents are planning agents since our future-directed projects provide a framework within which our practical thinking proceeds. Although the planning framework is not strictly necessary for the agency, it is a “deeply entrenched framework for us […] and one we may well not have the capacity to change at will” (Bratman 2018: 15). As such, planning agency is central to understanding human action.
For instance, Rawls (1971: 358) says that “a person is a human life lived according to a plan”. However, it is questionable whether people can choose a total life plan. As Heyd and Miller (2010) plausibly point out, this view presupposes that a life plan can be differentiated from the identity of the person whose plan it is; but our long-term plans often affect what kind of person we become by affecting the formation of new desires and preferences. Therefore, I take it that individuals can have several long-term plans concerning important dimensions of life such as profession, family, health, finances, which they pursue and revise in the course of their lives instead of having a total life plan.
Importantly, epistemically valid or justified beliefs should not be conflated with knowledge (Cf. Gettier 1963).
Not all epistemically valid expectations are also normatively valid, but all normatively valid expectations must be epistemically valid too. I take this point from Meyer and Sanklecha (2014).
Brown (2012) plausibly points out that to cause significant disruption in individual lives, it suffices that institutional revisions are large-scale even if they are not as frequent as Buchanan claims they are. The sheer scale of change suffices to raise issues of justice (e.g. systemic changes in response to the 2008 economic crisis). Arguably, the Fourth Industrial Revolution may increase the frequency of changes too, hence raising an even broader set of justice-related issues.
So defined, the harm resulting from frustrated legitimate expectations is different from the harm resulting from rights violations. Legitimate expectations are often characterized as “tacit understandings” generated by particular long-standing policies or provisions. As such, they constitute a distinctive form of entitlement that is to be differentiated from legally recognized rights (Brown 2011). The distinction between legitimate expectations and rights merits a treatment on its own.
Note that the frustration of legitimate expectations for the sake of justice makes a weaker pro tanto case for compensation compared to the frustration of the expectations for the sake of increasing efficiency. I will come back to this.
Some go as far as to claim that even frustration of illegitimate expectations may need to be compensated as it may be necessary for institutional reforms to take place. Namely, the frustration of expectations may make people resist change which, in turn, can make institutional attempts to introduce those changes futile and unfeasible (Trebilcock 2015).
According to Goodin (1995: 195–198), the main justification of welfare benefits is that they serve as protection from rapidly changing socio-economic conditions. The knowledge that such benefits are available even if they may never be claimed, stabilizes expectations and, thus, allows people to make and proceed with their plans.
Indeed, legal scholars point out that the system adaptability is part and parcel of the modern conception of legal certainty. Accordingly, the principle of legal certainty does not protect persons against all forms of legal changes, but only against the unpredictable ones (Popelier 2017).
Here I am focusing on those institutions that are legitimated through democratic authorization. For the discussion of the relationship between legitimacy and democracy see Buchanan (2002).
It may be informative to differentiate between the first-order legitimate expectations that concern particular policies and the second-order expectations that concern the legal system as a whole. The expectation of stability would be second-order in this sense.
Automation is only a part of a broader trend of the transformation of the labor market in terms of the growth of a contingent workforce via, for instance, the introduction of the gig economy and outsourcing, with much less job security. I do not consider these other sources of disruption here.
The term “technological unemployment” was coined by the economist John Maynard Keynes who argued that it may occur “due to our discovery of means of economising the use of labour outrunning the pace at which we can find new uses for labour” (Goos 2018: 366).
Although, Susskind (2022) argues that the previous economic models were predicting positive impacts of technology on the labour market based on underestimating the capabilities of technology. By contrast, the newer economic models that do take into account the great developmental potential of digital technology are giving much more pessimistic predictions.
This is not to say that those possessing low skills will not be hit hard, for, even if they preserve their job or can find another one, it is likely that their wages will keep stagnating or even decreasing if more people start competing for these jobs. I take this point from Loi (2015).
It is an empirical matter of when non-routine tasks may become “automatable” too (Goos 2018). In fact, Susskind (2022) points out that this is already happening in some areas such as medical diagnostics and self-driving cars. He argues that such developments could not have been predicted by the earlier economic models since the models presupposed an outdated understanding of how machines work (i.e. by following explicit instructions from humans). But AI is not so much about doing routine tasks by following human instructions as it is about learning on its own by recognizing and reproducing patterns.
This points to another immense problem – namely, the problem of the unequal distribution of the benefits of technological innovations both at the domestic and international levels. Internationally, the increasing employment of AI for labor purposes may reverse the trend of decreasing global inequality by moving jobs away from developing countries since computers at home can substitute cheap labor abroad (Schwab 2017). Distributive justice concerns arise not only regarding income inequality, but also regarding the uneven distribution of the risk of technological unemployment across the population (Parr 2022).
For instance, Rawls (1971) says that the choice of profession is a life plan. I understand professions as including a relatively stable set of skills. To say that someone has a profession means that the person has invested time and effort to acquire and exercise those skills. On this interpretation, professions are not limited to skills acquired through formal education, but also include skills developed through practicing, such as, for instance, various forms of crafts.
The reason for this is that autonomy as the capacity to develop and pursue plans is a unified capacity that is affected by things that are happening in different spheres of our lives. Those who do not have the opportunity to exercise this ability when performing their job tasks are at risk of leading less autonomous lives overall (Schwartz 1982).
It goes without saying that under the present conditions many people have very limited choices when it comes to professions for the reasons that are often beyond their control.
Note that the very term “upskilling” presupposes a value judgment that the new skills that may be needed are in some sense more valuable than those people have had. However, it is not clear that, for instance, a clockmaker who used to have a set of complex mechanical skills but now needs to retrain to take orders on a webshop interface to replace automatically produced digital watches will acquire valuable, higher skills. On the contrary, one may think that this amounts to downskilling rather than “upskilling”. Hence, acquiring certain digital skills does not necessarily amount to “upskilling”. I owe this point to Daniel Antal.
This concern is often overlooked. Instead, Luddites are typically portrayed as committing the so-called Luddite fallacy, that is, assuming that the amount of work is fixed and the more machines overtake it, the less there is for humans to do (Danaher 2019: 36).
The same goes for the collective level: significant resources may also be needed to keep retraining a large number of people.
I owe these examples to the anonymous reviewer.
I am grateful to an anonymous reviewer for pointing this out.
Danaher (2019) makes a similar point. While this leads him to conclude that humans should just drop this race, I derive different normative conclusions. I will come back to this in the next section.
Their analysis was limited to the US.
There are other, more provocative proposals that if humans cannot keep up with technology through education, perhaps they can do so through bioenhancement (Loi 2015). I have no space to engage with this version of Pandora’s box here.
Krugman (2013) points out that during the First Industrial Revolution workers who were hurt the most were those who had acquired valuable skills with much effort only to find them redundant and devalued.
Devaluing skills people took time and effort to acquire surely affects their self-respect, but this is a too large topic to address here.
I thank anonymous reviewer for drawing my attention to this.
The complementing process is happening in parallel with the substituting one (i.e. machines replacing human labour). Whether technological unemployment will take place depends on which of the two trends will overtake (Susskind 2022).
Lazar (2022) makes a similar point concerning the search and recommendation algorithms; namely, that those who design them exercise power over those who are affected by them.
This has clear implications for the freedom of occupation that many liberals hold dearly.
For instance, Danaher (2019: 119–120) argues that digital technologies undermine human autonomy by limiting the range of options to choose from and the capacity to make such choices independently from the interference (and domination) by others, but that they do not undermine the ability to plan.
To this, one may add that in light of such developments, deciding to enter those professions that are at a higher risk of being automated amounts to expensive taste rather than any kind of entitlement. To be sure, expectations to stay in a profession that is known to be at a high risk of automation would not be epistemically valid, and as such not legitimate; therefore, the frustration of such expectation would not ground a just claim for compensation. That said, there may be another issue involved here: I argued above that increasing deployment of digital technologies leads to a diminishing range of available professions people can choose from. Hence, those who may want to enter professions that are at a high risk of automation may have a claim on the grounds that their freedom of occupation has been significantly interfered with. I do not have space to further develop this point. I am grateful to an anonymous reviewer for raising this question.
One reactive proposal that may be resistant to this objection is universal basic income (UBI). Indeed, UBI is gaining more and more traction in these debates (Danaher 2019). However, to constitute a policy response that is adequate according to standards developed here (i.e. securing the individual ability to make and pursue long-term plans), UBI would need to be robustly basic.
A similar point has been made in Santoni de Sio, Almeida, and van den Hoven (2021).
Moreover, digital automation is largely driven by market logic and the imperative to cut costs of production, rather than by the desire to improve the position of the worst off. If anything, we could see earlier that automation leads to a growing polarization in jobs and wages. Therefore, legitimate expectations in the sphere of work are frustrated on the grounds of increasing efficiency rather than justice.
It is often ignored that deployment of digital technology has significant environmental impact since it consumes vast amount of energy and also produces waste that is hard to recycle (Müller 2021).
To be sure, the decreased capacity of state institutions to deal with social challenges is not due to technological developments only, but it is greatly affected by other processes as well, most of which is globalization. For an empirical study see Mair (2013), for a normative account see Ronzoni (2013).
Much of liberal political thought focuses on the justifiability of the coercive imposition of laws and decisions by the government while assuming that governmental institutions and bodies are able to do this. However, the trend of globalization and now technological development challenges this assumption to a great extent. To be sure, there could be an increase in state power if governments use technology for dubious purposes such as surveillance. This is too large a question to go into here. For seminal classical works on political legitimacy see Hobbes (1968), Hume (1993) and Locke (1698) and for more contemporary ones, see, for instance, Simmons (2000).
References
Acemoglu, Daron, and David Autor. 2012. What does human capital do? A review of Goldin and Katz’s ‘the race between education and technology.’ Journal of Economic Literature 50(2): 426–463.
Arneson, Richard J. 1987. Meaningful work and market socialism. Ethics 97(3): 517–545. https://doi.org/10.1086/292864.
Autor, David H. 2015. Why are there still so many jobs? The history and future of workplace automation. Journal of Economic Perspectives 29(3): 3–30. https://doi.org/10.1257/jep.29.3.3.
Autor, David, and David Dorn. 2013. “How Technology Wrecks the Middle Class.” Opinionator, The New York Times (Blog).
Benn, Claire, and Seth Lazar. 2022. What’s Wrong with automated influence. Canadian Journal of Philosophy 52(1): 125–148. https://doi.org/10.1017/can.2021.23.
Bratman, Michael E. 2018. Agency, Time, and Sociality. In Planning, Time, and Self-Governance: Essays in Practical Rationality, ed. M.E. Bratman. Oxford University Press. https://doi.org/10.1093/oso/9780190867850.003.0005.
Brown, Alexander. 2011. Justifying compensation for frustrated legitimate expectations. Law and Philosophy 30(6): 699–728. https://doi.org/10.1007/s10982-011-9108-z.
Brown, Alexander. 2012. Rawls, Buchanan, and the Legal doctrine of legitimate expectations. Social Theory and Practice 38(4): 617–644.
Buchanan, Allen. 1975. Distributive justice and legitimate expectations. Philosophical Studies 28(6): 419–425. https://doi.org/10.1007/BF00372903.
Buchanan, Allen. 2002. Political legitimacy and democracy. Ethics 112(4): 689–719.
Colvin, Geoff. 2015. Humans Are Underrated. Fortune (Blog).
Danaher, John. 2019. Automation and Utopia: Human Flourishing in a World without Work. Harvard University Press. https://doi.org/10.4159/9780674242203.
Dobbs, Richard, James Manyika, and Jonathan Woetzel. 2015. The Four Global Forces Breaking All the Trends | McKinsey. The Four Global Forces Breaking All the Trends (Blog).
Feinberg, Joel. 1973. Duty and obligation in the non-ideal world. Journal of Philosophy 70(9): 263–275.
Frey, Carl Benedikt, and Michael A. Osborne. 2017. The future of employment: How susceptible are jobs to computerisation. Technological Forecasting and Social Change 114: 254–280.
Gettier, Edmund L. 1963. Is justified true belief knowledge? Analysis 23(6): 121–123. https://doi.org/10.2307/3326922.
Gheaus, Anca, and Lisa Herzog. 2016. The goods of work (Other Than Money!). Journal of Social Philosophy 47(1): 70–89. https://doi.org/10.1111/josp.12140.
Goldin, Claudia, and Lawrence F. Katz. 2007. The race between education and technology: The evolution of U.S. educational wage differentials, 1890 to 2005. w12984. National Bureau of Economic Research. https://doi.org/10.3386/w12984.
Goodin, Robert E. 1995. Utilitarianism as a Public Philosophy. Cambridge University Press.
Goos, Maarten. 2018. The impact of technological progress on labour markets: Policy challenges. Oxford Review of Economic Policy 34(3): 362–375. https://doi.org/10.1093/oxrep/gry002.
Green, Fergus. 2017. Legitimate expectations, legal transitions, and wide reflective equilibrium. Moral Philosophy and Politics 4(2): 177–205. https://doi.org/10.1515/mopp-2016-0029.
Hedman, Carl G. 1989. Luddites, Hippies and Robots: Automation and the possibility of resistance. Prometheus 7(2): 273–291. https://doi.org/10.1080/08109028908629074.
Heyd, David, and Franklin G. Miller. 2010. Life plans: Do They give meaning to our lives? The Monist 93(1): 17–37.
Hobbes, Thomas; 1968. Leviathan. Edited by C. B. Macpherson. Harmondsworth: Penguin Classics.
Hume, David. 1993. Essays Moral, Political, and Literary. Oxford University Press.
Hurka, Thomas. 1987. Why value autonomy? Social Theory and Practice 13(3): 361–382.
Kallis, Giorgos. 2018. Degrowth. Newcastle upon Tyne: Agenda Publishing.
Krugman, Paul. 2013. Sympathy for the Luddites. The New York Times.
Laitinen, Arto, and Otto Sahlgren. 2021. AI systems and respect for human autonomy. Frontiers in Artificial Intelligence. https://doi.org/10.3389/frai.2021.705164.
Lazar, Seth. 2022. Power and AI: Nature and Justification. In The Oxford Handbook of AI Governance, ed. Justin B. Bullock, Yu-Che. Chen, Johannes Himmelreich, Valerie M. Hudson, Anton Korinek, Matthew M. Young, and Baobao Zhang. Oxford University Press.
Locke, John. 1698. Two Treaties of Government. 3rd ed. London: Awnsham and John Churchill.
Loi, Michele. 2015. Technological unemployment and human disenhancement. Ethics and Information Technology 17(3): 201–210. https://doi.org/10.1007/s10676-015-9375-8.
Mair, Peter. 2013. Ruling the Void: The Hollowing of Western Democracy. Verso Books.
Meyer, Lukas H., and Pranay Sanklecha. 2011. Individual expectations and climate justice. Analyse & Kritik 33(2): 449–472. https://doi.org/10.1515/auk-2011-0208.
Meyer, Lukas H., and Pranay Sanklecha. 2014. How legitimate expectations matter in climate justice. Politics, Philosophy & Economics 13(4): 369–393. https://doi.org/10.1177/1470594X14541522.
Müller, Vincent C. 2021. Ethics of Artificial Intelligence and Robotics. In The Stanford Encyclopedia of Philosophy, ed. Edward N. Zalta, Summer 2021. Metaphysics Research Lab, Stanford University.
Parr, Tom. 2022. Automation, unemployment, and insurance. Ethics and Information Technology 24(3): 24. https://doi.org/10.1007/s10676-022-09639-w.
Popelier, Patricia. 2017. The Shifting Meaning of Legal Certainty in Comparative and Transnational Law. In The Shifting Meaning of Legal Certainty in Comparative and Transnational Law, ed. Mark Fenwick, Mathias Siems, and Stefan Wrbka, 29–54. London: Hart Publishing Ltd.
Räikkä, Juha. 2014. Social Justice in Practice: Questions in Ethics and Political Philosophy. Studies in Applied Philosophy, Epistemology and Rational Ethics, Volume 14. Cham: Springer.
Rawls, John. 1971. A Theory of Justice. Harvard University Press.
Raz, Joseph. 1988. The Morality of Freedom. Clarendon Press.
Ronzoni, Miriam. 2013. For (Some) Political and Institutional Cosmopolitanism, (Even If) Against Moral Cosmopolitanism. In Cosmopolitanism versus Non-Cosmopolitanism: Critiques, Defenses, Reconceptualizations, ed. Gillian Brock, Oxford University Press.
Schwab, Klaus. 2017. The Fourth Industrial Revolution. Illustrated. New York: Currency.
Schwartz, Adina. 1982. Meaningful work. Ethics 92(4): 634–646.
Simmons, A. John. 2000. Justification and Legitimacy: Essays on Rights and Obligations. Cambridge: Cambridge University Press.
Simmons, A. John. 2010. Ideal and nonideal theory. Philosophy and Public Affairs 38(1): 5–36.
Santoni de Sio, Filippo, Txai Almeida, and Jeroen van den Hoven. 2021. The Future of Work: Freedom, Justice and Capital in the Age of Artificial Intelligence. Critical Review of International Social and Political Philosophy 0(0): 1–25. https://doi.org/10.1080/13698230.2021.2008204
Susskind, Daniel. 2022. Technological Unemployment. In The Oxford Handbook of AI Governance, ed. Justin B. Bullock, Yu-Che Chen, Johannes Himmelreich, Valerie M. Hudson, Anton Korinek, Matthew M. Young, and Baobao Zhang, online edition. Oxford University Press. https://doi.org/10.1093/oxfordhb/9780197579329.013.42.
Szende, Jennifer. 2020. Planning in the Void: Autonomy Amid Pandemic Constraints. Apa Newsletter on Feminism and Philosophy 20(1): 26–28.
Trebilcock, Michael J. 2015. Dealing with Losers: The Political Economy of Policy Transitions. Reprint. New York: Oxford University Press.
Willcocks, Leslie. 2020. Robo-Apocalypse Cancelled? Reframing the automation and future of work debate. Journal of Information Technology 35(4): 286–302. https://doi.org/10.1177/0268396220925830.
Acknowledgements
I greatly benefitted from discussing the first draft of the manuscript with W. Jared Parmer, Filippo Santoni de Sio and other participants of the “AI and the Future of Work” workshop organized at TU Delft. I would also like to thank Daniel Antal and two anonymous reviewers for their insightful comments on the previous version of this article.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
There are no competing interests to disclose.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Belic, J. Institutions, Automation, and Legitimate Expectations. J Ethics (2023). https://doi.org/10.1007/s10892-023-09440-x
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s10892-023-09440-x