Meaningful Work and Achievement in Increasingly Automated Workplaces

Abstract

As automating technologies are increasingly integrated into workplaces, one concern is that many of the human workers who remain will be relegated to more dull and less positively impactful work. This paper considers two rival theories of meaningful work that might be used to evaluate particular implementations of automation. The first is achievementism, which says that work that culminates in achievements to workers’ credit is especially meaningful; the other is the practice view, which says that work that takes the form of an open-ended practice is especially meaningful. Of the two, the practice view is the better tool for assessing the future of meaningful work, because achievementism is explanatorily inadequate in two ways. Moreover, the practice view can explain why the most meaningful forms of work cannot be automated. A procedure can only be automated as long as the steps are defined, whereas, in such work, which steps to take is indefinitely open to redefinition. This reveals the real threat to meaningful work to be a political-economic one. The relevant ethical questions there have to do with how much creative control workers retain in crafting their own jobs, when those jobs involve meaningful work in collaboration with machines; and how to liberate workers from jobs that do not involve meaningful work, which should be left to machines anyway.

Especially in light of rapid and sometimes stunning advancements in artificial intelligence and robotics, one concern one might have is that automating technologies tend to displace workers into more dull and less positively impactful work. Perhaps with the exception of the privileged few engineers who design such technology in the first place, and of the executives who direct the deployment of that technology, the rest of us might be increasingly relegated to maintaining, cleaning up after, and overseeing automating technologies that actually produce the goods and services people value. Moreover, it may be that this displacement is coming for (almost) all of us, even those whose work we are not accustomed to thinking of as vulnerable to automation, such as teachers, nurses, content-creators, and researchers.

Of course, there is a lively debate about the effects of automation in the workplace – what the effects on productivity and (the levels and structure of) employment actually are, and so on. So it may be that this concern is founded on an unwarranted empirical presupposition about how work changes in response to automation. Nevertheless, there are two normative questions in the background here: Would such displacement be bad (all else equal) were it to occur? And, if it would be bad, in virtue of what would it be bad? Here is a natural line of thought that answers both questions. All else equal, when a worker is displaced by automation into more dull and less positively impactful work, that is bad because she will be left with less meaningful work than before. The idea that meaningful activity is both positively impactful and subjectively engaging enjoys fairly widespread acceptance among philosophers working on the topic,¹ and, even among alternate views, one or the other feature (positive impact or subjective engagement) is virtually ubiquitous as a necessary feature.²

Still, we should entertain second thoughts about this concern. When a worker shifts from herself producing some valuable good or service to maintaining, cleaning up after, or overseeing some automating technology that produces the same good or service, is she thereby relegated to more dull and less positively impactful – less meaningful – work? It is not obvious that she is. For instance, why should maintaining some complex piece of machinery be more dull than doing, by hand, whatever job it is supposed to do? In plenty of cases, actually providing some good or service can be repetitive and simple – think about performing a simple spot-weld on thousands of pieces of metal a day, or taking the temperatures of every patient on a hospital floor – whereas maintaining the machine that does the job instead can be varied and complex. Or, when a worker is responsible for overseeing some piece of technology that produces much more of some good, and produces it more efficiently, than she could herself, why should her oversight be less positively impactful than if she produced that good herself? It is not obvious how to measure, and then compare, the impact of the work in each case.

These questions will turn on the details, and would need to be answered via hybrid empirical and normative investigation – because not only would we need to know what some particular case of displacement-by-automation looks like, but we would need to exercise our normative judgment concerning the meaningfulness of the resultant work.

At this closer level of analysis, however, further ethical theorizing may be brought to bear to refine our normative judgment. One possibility is that certain kinds of subjective engagement, and certain kinds of positive impact, are more meaningful than others. If that is right, our judgments must take into consideration not only how engaging or impactful the displaced worker’s work is now, compared to the work she did before; they must also take into consideration the character of her engagement or impact now, compared to before.

What I will call achievementism about meaningful activities offers one such further refinement. In brief, achievements are challenging activities that conclude in some outcome, where the activity substantially and non-luckily caused that outcome. And achievementism is the view that, all else equal, achievements are especially meaningful activities.³ Achievementism holds promise as a tool for assessing meaningful work after displacement by automation, and it helps to precisify the initial worry with which I began. Nevertheless, I argue, achievementism is not the tool we should use, and the initial worry is therefore, perhaps, misguided.

Here is the plan for the paper. In Sect. 1, I spell out what achievements are, and what achievementism is. Then I motivate using it for assessing meaningful work after displacement by automation. In Sect. 2, I introduce and motivate my conception of inexhaustibly meaningful activities as practices, which are non-completable activities in which the standards for pursuit are worked out in the practice itself. These activities capture core cases of intuitively highly meaningful activities, and explain the derivative but exhaustible meaningfulness that some activities can have. In Sect. 3, I argue that this practice view is superior to achievementism as a tool for assessing meaningful work. Properly speaking, inexhaustibly meaningful work cannot be achieved at all, and this is the first flaw of achievementism. And, even in cases most friendly to achievementism, achieving ends does not make that work meaningful. Rather, what explains why that work is meaningful are the inexhaustibly meaningful practices of which it is a part. This is the second flaw of achievementism.

I mentioned at the outset that the initial worry is, perhaps, founded on shaky empirical grounds. But if the argument of this paper is correct, it is also probably founded on a misconception of meaningful work. Thus, I wrap up in Sect. 4 with a discussion of what automation does show about meaningful work.

1 Achievements, Achievementism, and the Threat of Automation

1.1 What Achievements Are

For achievementism to be a substantial refinement of a theory of meaningful activity, achievements cannot simply be activities with positive impacts. But this distinction has intuitive pull. On accounts of what achievements are, it is standardly emphasized that, when a person achieves something, her activity non-luckily and substantially caused the outcome, and was in some way challenging.⁴ Positive impacts, by comparison, can be brought about incompetently, accidentally, or with minimal contribution by an agent; and bringing about a positive impact need not be challenging for the agent who does so. For example, if I accidentally prevent you from burning your fingers when, as you reach to pick up a slice of pizza, I distract you with a funny story, my telling the story has a positive impact, but my saving your fingers is not an achievement because it was entirely by accident. Or if it is no hardship at all for me to pay your water bill for you when you are down and out, my doing so is also not an achievement because it is not challenging in any way for me to do it.

Now, and this is going to be important, the only candidates for achievement are completable activities, activities that conclude or terminate upon success.⁵ Most ready-to-hand examples of activities are completable: walking to the store, writing a book on Kant’s Groundwork, nursing a loved one back to health, etc. For such activities, when we succeed, we are done. Another way to identify completability is to check whether, upon succeeding, it is possible to continue to do the activity or only do it again. If one can only do it again, it is a completable task. For example, when I have successfully walked to the store, I am at the store and done; I simply cannot continue to walk to the store, though I may do it again. When we successfully complete such activities, and doing so is challenging, and substantially and non-luckily concludes in the outcomes that are their point, they are achievements. It can be an achievement to walk to the store (if, for example, you are recovering from a serious injury), to finish that manuscript, and to support your partner until she is well again.

By contrast, other activities do not conclude or terminate upon success.⁶ They are not completable tasks. For these activities, when we succeed, we need not be done; and, upon success, it is possible to continue to do them. To see the contrast, compare walking to the store (completable) with going for a stroll (not completable); or writing a book on Kant’s Groundwork (completable) with doing philosophy (not completable); or nursing a loved one back to health (completable) with caring for her more generally (not completable). When I am going for an aimless stroll (rather than walking somewhere specific, such as the store), I can succeed in this even while not being done, and nothing precludes me from continuing successfully to do it.

Moreover, non-completable activities are not candidates for achievements. The reason for this is simply that there is nothing to point to as the achievement. This is so even when the activity is challenging, and the agent is substantially and non-luckily causing things to happen as a result of her activity. Imagine, by way of example, that I am recovering from a nasty injury, but one day I manage to go for a stroll. It is challenging, surely, and we may suppose that it is substantially and non-luckily helping me to return to my earlier mobility. But what am I supposed to have achieved? This question gets no proper answer. The natural response, of course, is to point to the actual walk I took, perhaps emphasizing how much longer it was than anything I had walked since my injury or that I did not lean as heavily on my crutches. But to answer in this way is to point to some particular stretch of the non-completable activity I was doing, some part of it that was completed or concluded. It is to identify the completable activity that, in doing the non-completable activity of going for a stroll, I also did.

This feature is not always adequately emphasized in analyses of achievements. For example, whereas Gwen Bradford characterizes achievements as processes (activities) culminating in a product (an outcome), she at times wants to allow that the product can be ‘an ongoing state of affairs’ such as being a philosopher – viz., a non-completable activity.⁷ But this conflates achievements with challenging, successful activity. It should come as no surprise that I am happy to agree that being a philosopher is challenging (to do well). However, there is nothing to point to as the achievement I have achieved. This is precisely because a key part of achievements is that they culminate in some product. They conclude or are complete with the production of some outcome. And yet, by Bradford’s own lights, being a philosopher does not culminate in anything; it just goes on.

This is not merely a semantic dispute. The completability of achievements plays a key role in making achievementism a promising tool for evaluating meaningful work after displacement by automation. I turn to that now.

1.2 Achievementism as a Tool for Assessing Meaningful Work

Achievements are challenging activities that substantially and non-luckily conclude in some outcome. Achievementism is the view that, all else equal, achievements are especially meaningful activities. Here, then, is the refinement that achievementism offers. When appended to the view that meaningful activity is both subjectively engaging and positively impactful, achievementism says that certain kinds of subjective engagement, and of positive impact, are more meaningful than others (all else equal). In particular, being challenging for the agent is an especially meaningful form of subjective engagement – stimulating, enlivening, and demanding of her cognitive and conative resources. And the especially meaningful form of positive impact will be those outputs that are substantially caused, in a non-lucky way, by the conclusion of what she is doing.

Why should we take this offer? Let me now motivate adopting achievementism to assess meaningful work after displacement by automation. The idea is that, when a worker is displaced by automation, we should not only consider how subjectively engaging or positively impactful her work now is compared to before, but what opportunities for achievement in her work she now has compared to before.⁸ This idea promises to recover an aspect of the concern with which we began, even after entertaining the second thoughts we did. When a worker, who previously produced some valuable goods or services herself, now maintains, cleans up after, or oversees some automating technology that now produces those goods or services itself, we said it was not obvious that her new work would be less subjectively engaging, or less positively impactful, than the work she did before. But what is plausible is that the valuable goods or services produced no longer redound to her credit as achievements of hers; for now she is, basically, a helper for the machine that produced them.

Our account of achievements can explain why this is a plausible thought to have, and achievementism would explain why it should concern us. First, achievements are challenging activities. Whether or not some case of displacement-by-automation renders a worker’s work less challenging than it was before will depend a great deal on the details of that case. As with subjective engagement more generally, this can vary: There will, for example, be cases in which maintaining a piece of complex machinery that produces some good is more challenging for the worker than producing that good herself. Still, achievementism tells us, specifically, one kind of subjective engagement we should be looking for and assessing. Second, achievements involve substantial, non-lucky causal contribution. Here, again, the details of the case will matter – for example, how active and involved the human worker’s oversight of the automating technology is, compared with how active and involved producing the goods or services herself was. Still, achievementism tells us, specifically, to look and assess the worker’s substantial and non-lucky causal contribution to the positive impacts that are relevant – which, in most cases, will be the valuable goods or services that are produced.

Third, achievements are completable activities. When assessing a case of displacement by automation, achievementism says that it matters whether the work the worker did prior concluded in positive impacts, and whether her work now does. A clear threat to the meaningfulness of her work will be cases in which displacement eliminates from her work activities that conclude in positive impacts, where these activities were there before. Strikingly, the kinds of displacement we are considering look precisely to be that way. We are considering cases in which human workers initially produce valuable goods or services themselves, where this production will in familiar cases be the outcome of completable activities: diagnosing an illness, cooking a meal, making a coat, and so on. After being displaced, the worker maintains, cleans up after, or oversees an automating technology that produces those goods or services itself. Thus, the technology now seems to perform the particular completable activity that she performed before, which concludes in the positive impact that matters. And, importantly, the activity she is now doing – maintenance, cleaning, or oversight – is a non-completable activity that just goes on. It is not something she can complete, hence not something she can achieve.

It thus appears that the completability of achievements is where achievementism gets its firmest grip on the concern. Because only completable activities can amount to achievements, completability gives us a categorical criterion to assess a worker’s opportunities for achievement after being displaced by automation. Notice that this criterion does not turn on empirical details in the way the other criteria for achievements do, but rather on the conceptual question of what the worker is doing before and after displacement. When I gave initial voice to the concern about automation, I gave it by drawing a contrast in the kinds of work being done that, we can now appreciate, relied on the distinction between completable and non-completable activities.⁹

To buttress this point, let me show how completability as a categorical criterion of achievement clarifies John Danaher and Sven Nyholm’s critique of automation in the workplace (2021). In brief, they assume that achievementism is true, and then evaluate how worker displacement by automating technology can threaten opportunities for achievement in work, and thereby threaten (or at least diminish) the meaningfulness of the work those workers do in the aftermath. The kind of displacement that most interests them is what they call ‘collaborative displacement’, which broadly tracks the displacement I have been emphasizing, and takes three forms: human works remain on the job, but now act as supervisory, maintenance, or order-following collaborators with the technology (2021: 229). However, because they are not mindful of the fact that only completable activities can amount to achievements, their critiques are less clear than they could be.

Here are a couple of key argumentative moments in their discussion. Concerning maintenance collaboration with technology, they say:

Where once upon a time humans could be responsible for producing valuable commodifiable outputs [which can be either goods or services]... when they shift to the maintenance role they necessarily take a peripheral role in the production of the valuable outputs. They keep the wheels of industry turning, but they do not play a direct role in the valuable activities (2021: 232, emphasis mine).

Now, playing ‘a direct role’ in valuable activities is somewhat opaque, but is perhaps illuminated by the analogy they go on to make between a musical performer and her road crew, who set up, break down, and repair all the equipment she needs to perform. Here, they say that ‘the real value’ of the activities involved is the performance itself, and that the road crew’s work cannot be more valuable than the musician’s performance (ibid.). But this does not show that, when a worker has been displaced from a ‘frontstage’ role to a ‘backstage’ role of maintenance, her work is any less valuable for it. It does plausibly show that the kind of value her work has has changed, to something more instrumental (no pun intended) than before. But that is compatible with her work having as positive an impact as it did before; it is just that the nature of the impact is different. Why should this diminish opportunities for achievements in work? The answer is not plausibly that maintenance work is less challenging, or that it is less substantially and non-luckily contributory to positive impacts. More plausibly, the answer will appeal to the fact that the maintenance workers’ (the road crew’s) activities do not conclude in the valuable product that matters – the goods or services rendered (the concert performed). They do not engage in activities that are complete upon or conclude in the relevant outcome, and so that outcome cannot amount to an achievement of theirs.

Further on, they discuss supervisory collaboration – in which human workers direct and oversee work done by automating technology – and also argue that it threatens opportunities for achievement in work. As an illustration of supervisory collaboration, they point to the development of AlphaGo, an AI capable of beating high-ranking professional human Go players such as Lee Sedol, which wins by drawing on self-specified learning and search algorithms. Importantly, while AlphaGo has a team of Google-employed engineers behind it, who ‘supervise’ it, its tactics are largely not decided by them, nor are they privy to the decision-making underlying such tactics. According to Danaher and Nyholm, these engineers “were not, in any meaningful sense, responsible for its specific successful moves within [any particular] game” (2021: 233). But why would they need to be? The explanation they offer is as follows:

If costly commitment [i.e., how challenging the outcome is to bring about] is one of the key variables for assessing the value of an achievement, it would seem to follow that automation reduces the value of workplace related achievements for workers. The designers of Alpha Go would…appear to be a case in point. They did not have to do the hard work of figuring out the optimal strategies to play in the game, nor did they have to code these strategies into the computer (2021: 233).

So the explanation they offer seems to be this: Workers displaced into supervisory roles of automating technology are left with work that is less challenging than before. Because achievements are challenging activities, the threat is that their work will be insufficiently challenging to constitute an achievement. But, again, this is not all that obvious, and (turning their argument on its head) the designers of Alpha Go would appear to be a case in point: Designing an AI capable of beating one of the best Go players in the world is no mean feat. Danaher and Nyholm’s emphasis on what the designers of Alpha Go are not doing – thinking through and making plays against Lee Sedol – suggests, rather, a different explanation. Danaher and Nyholm seem interested more in the fact that the designers’ activities do not conclude in the product that matters, namely, the stunning victory itself. The designers did not beat Lee Sedol. In other words, they do not do activities that are complete with or conclude in the relevant outcome, and so that outcome cannot amount to an achievement of theirs.

I take the foregoing discussion to motivate using achievementism, coupled with a broader theory of meaningful activity as subjectively engaging and positively impactful, as a tool for assessing meaningful work after displacement by automation. Achievementism enjoys some intuitively plausibility, and promises leverage that the broader theory alone does not provide.

Nevertheless, in the remainder of this paper, I will argue that achievementism is false. Because much of my argument will turn on the completability of any activity we can achieve, I have taken pains to explain how this feature of achievements is key to the leverage achievementism promises.

2 The Meaningfulness of Pursuit and Progress: The Practice View

For a moment, let us set achievementism aside and consider an alternative picture of especially meaningful activities, the practice view. For our purposes, we are thinking of achievementism as a specification of the broader view that meaningful activities are subjectively engaging and positively impactful – because it specifies what kinds of subjective engagement and positive impact are (all else equal) more meaningful than others. We can think of the practice view similarly, but as a different specification. As we will see, achievementism and the practice view are somewhat at loggerheads, so we will need to consider what speaks in favor of using each to assess meaningful work. We have already seen what speaks in favor of achievementism. So, in what follows, I will spell out the practice view and motivate adopting it. Then, in Sect. 3, I will argue that the practice view is superior.

In brief, the practice view is that practices are especially meaningful activities. This involves two key commitments. The first is that especially meaningful activities are inexhaustibly meaningful activities – activities we can simply keep doing, and they will keep making our own lives meaningful by doing them.

Why should we care whether our meaningful activities are inexhaustible in this way? The answer comes in two parts. First, the activities we standardly think of as the most meaningful exhibit just this feature. The most ready-to-hand examples, such as caring for one another, or pursuing knowledge, or masterfully building things, are not activities we expect to run dry: It is not as though we expect their meaningfulness to one day be exhausted, necessitating our finding new activities that will imbue our lives with meaning going forward. Second, for the meaningful activities we do expect (or hope) to run dry someday – such as searching for more effective cancer treatments, or writing a groundbreaking work of philosophy, or shepherding our children into adulthood – an inexhaustible activity in the background usually explains their meaningfulness. For example, treating cancer is of a piece with caring for one another in the face of privation, and writing a groundbreaking work of philosophy is of a piece with searching for understanding. Moreover, failing to keep this inexhaustible, background activity in view makes one vulnerable to a host of well-known crises – think of the midlife crisis of a successful careerist, or the ‘all-for-nothing’ crisis of an unsuccessful one – and so is generally ill-advised. This suggests, in turn, that when there simply is no inexhaustible activity in the background, something has gone wrong.¹⁰ So there is some reason to think that the practice view is right that inexhaustibly meaningful activities are especially meaningful.

The second key commitment of the practice view is that inexhaustibly meaningful activities are practices. Practices are activities in which a person competently and indefinitely pursues goals that she refines and expands along the way.¹¹ This dynamic is perhaps most obvious in research and art – think about how research questions evolve over time, or how an artist cultivates, and even goes beyond, her own distinctive style. However, it is as much present in care work such as nursing or teaching, wherein the caregivers’ relationship to those they care for, and the standards internal to it, can deepen and become more rich over time. In all such cases, moreover, the activity has no termination point – it is not completable. Practices are thus essentially open-ended in two respects: They neither specify a termination point, nor afford standards at the outset. We define such standards in ongoing practice.

Why think that inexhaustibly meaningful activities are practices? The answer is that such activities are open-ended in precisely the way that practices essentially are. First, notice that inexhaustibly meaningful activities paradigmatically lack termination points, and so cannot be completed. After all, cultivating a distinctive artistic style, doing good philosophy, caring for a loved one, or developing excellence in a craft do, indeed, appear to be both inexhaustibly meaningful and non-completable. In being non-completable, they resemble practices.

But this is not the whole story, since an activity might not be completable yet be only exhaustibly meaningful. Consider, for example, going for enjoyable strolls, spending time with one’s friends, and tending to a garden. These activities cannot be completed, but seem only exhaustibly meaningful because they threaten to be, eventually, endlessly repetitive.¹²

That is why the second aspect of practices matters, namely, that we work out the standards for pursuing a practice in the pursuit itself. For theorists like Levy (2005) and Bradford (2022), it is intuitive that inexhaustibly meaningful activities also exhibit this feature. But let me provide some independent motivation.¹³

First, merely non-completable activities (going for aimless strolls, and so forth) seem to be missing the possibility of progress in their pursuit, of the sort that standards might provide. However, it does not seem that such standards can be merely given. Here is an illustration. Suppose that the rules of (classic) Sudoku, along with a finite variety of sophisticated solving techniques (such as the X-Wing, the Swordfish, etc.), are simply given. Now, we can take up the non-completable activity of solving Sudokus, and these given rules and techniques can secure the possibility of progress, but only up to a point: By learning to follow the rules and deploy the most sophisticated techniques as appropriate, one can solve progressively more difficult puzzles, and solve them more quickly. Eventually, however, one will simply master these rules and techniques, and there will be nothing left to do but continue to solve Sudokus as if by rote. Thus repetition still looms, delayed though it is.¹⁴

Second, consider how we actually go about doing inexhaustibly meaningful activities. Whereas things like being a good friend or an excellent nurse, or doing good philosophy, are paradigmatic cases of inexhaustibly meaningful activities, they are not activities that we can simply do straightaway. These activities need further, actionable specification that makes them amenable to familiar patterns of planning toward preliminary steps and means, and to reconciliation with the other plans we have. We need, in other words, some kind of actionable goal to pursue so as to perform them. However, such actionable goals will, in the best cases, amount to over-specifications of inexhaustibly meaningful activities.

Here are a couple of examples. Doing good philosophy would be over-specified by, say, having the goal of proving transcendental idealism; or, even more so, by having the goal of writing a paper that presents a few counterexamples to hedonism. Being a good friend to someone in particular would be over-specified by having the goal of supporting them in all their (worthy) professional endeavors. These would not be misspecifications – if transcendental idealism turns out to be true, for example, then proving it is perfectly of a piece with doing good philosophy – but rather cases wherein the specified goal screens off ways of doing the inexhaustibly meaningful activity in question. And yet, as I have said, doing good philosophy is under-specified in the sense that a person cannot take it up straightaway and formulate a plan for how to do it. Doing good philosophy, as such, has too little content to support anything like concrete steps or means to take in its pursuit.

So we end up with two sorts of problems. The first is that a merely non-completable activity, or an activity with merely given standards for pursuit, threatens to be endlessly repetitive and hence not inexhaustibly meaningful. The second is that inexhaustibly meaningful activities seem only actionable once they have been over-specified in some manner.

Appealing to the second aspect of practices’ open-endedness promises to solve these problems at a stroke. The open-endedness of practices refers not only to their non-completable character, but to the way the relevant standards of pursuit are worked out in pursuing the practice itself.¹⁵ More specifically: By being open to, and learning from, experience in pursuing an actionable but over-specified goal, we develop a richer understanding of the standards in light of which further actionable but over-specified goals should be specified going forward. Following Millgram (2008), I will mark this dynamic by saying that actionable goals associated with practices – such as doing philosophy – are specified widely over time: We specify, take up, and move beyond over-specified goals – such as concocting counter-examples to hedonism – balanced against a simultaneous specification of the standards this latter kind of goal must meet. We pay the price of over-specification for the benefit of actionability, but we keep the price from being prohibitive by relying on a more basic capacity to learn from experience.

To be sure, there is more to say here. Let me just mark a couple of reasons why. The first is simply that wide specification of goals, as a mode of practical thinking, is not very well understood.¹⁶ This means, among other things, that we do not have a solid basis for saying when, in general, an actionable goal amounts to an over-specification rather than a misspecification. Now, be that as it may, specification is observably a part of thinking about what to do. The lacuna here is evidently in philosophy, not life. Anyway, it is not clear that an account of specification, suitably general to satisfy philosophers working today, is in the offing. This is because the distinction between over-specification and misspecification is thoroughly situation-specific. Just to give an illustrative gesture: there are situations in which even going diving counts as a part of summiting a mountain, such that the former can constitute a step toward the latter, actionably specified. It can do so, for example, when one specifies the latter by adopting the goal of climbing the easy side of the mountain, which is only accessible via an underground lake.¹⁷

Second, if the moral to be drawn from my Sudoku example is taken seriously, then wide specification ameliorates the threat of repetition only as long as the standards involved can be refined and expanded further. This means, in turn, that it only really avoids repetition if this process can proceed indefinitely. And this might strike some as simply incredible: Can we really believe that the standards involved in being a good friend, or doing good philosophy, or making society more just, or creating singular art admit of indefinite modification? Aren’t the real standards for such activities, ultimately, definite and subject to our discovery? This issue points to deeper disagreements that crop up, in various ways, in metaethics, aesthetics, the philosophy of science, and metaphilosophy. Here, let me just provide an illustrative gesture as to why the implication is not simply incredible. Friendship develops over time, and changes each friend’s interests, personality, and so on; these changes, in turn, change the relationship between these friends; and this interactive dynamic at least partly grounds whatever standards each friend should meet vis-à-vis the other at any particular time. There’s thus nothing like a stable, complete set of standards available to either friend; and so the insistence that, ultimately, some such set of standards is there to be known looks more like a bet than a principled avowal.

Setting these issues aside, the second aspect of practices – what I have here analyzed as indefinite wide specification – promises to solve two important problems in one go. We noticed that merely non-completable activities threatened to eventually be endlessly repetitive, and thus not inexhaustibly meaningful. Wide specification avoids this because the standards of pursuit are indefinitely subject to refinement as we learn from experience. And we noticed that inexhaustibly meaningful activities are not activities we can, as planning agents, do straightaway. Wide specification solves this problem by articulating actionable but over-specified goals. Thus, this aspect of practices promises to capture how inexhaustibly meaningful activities are inoculated from repetition, and how they are pursued by agents like us.

To sum up, the practice view privileges inexhaustibly meaningful activities as those that are especially meaningful, and it promises to capture their inexhaustibility by appeal to the essentially open-ended character of practices. This maneuver enjoys intuitive support from the observation that paradigmatically especially meaningful activities – such as caring for one another, or doing philosophy, or developing excellence in a craft – look to be inexhaustibly meaningful, and to exhibit this open-endedness. And it enjoys further support from consideration of how planning agents can stave off repetition and make progress in the pursuit of inexhaustibly meaningful activities.

Finally, the practice view couples naturally with the broader view of meaningful activity as subjectively engaging and positively impactful, because it specifies what kinds of subjective engagement and positive impact are (all else equal) more meaningful than others. In particular, positively impactful activities that are non-completable are more meaningful than those that are not; and a continued focus on, and wide specification in the pursuit of, those very activities is a more meaningful kind of subjective engagement than an excessive focus on completable activities, or the mere acceptance of actionable goals and standards of pursuit.

3 Displacement by Automation and Meaningful Work

Which of these views – achievementism or the practice view – should we adopt to assess meaningful work after displacement by automation? First, we will consider inexhaustibly meaningful work, then we will turn to exhaustibly meaningful work. I argue that we should adopt the practice view over achievementism on both counts.

To clear the ground a bit: Work is not the same thing as a job. At a minimum, work is productive activity that is necessary in a broad sense – which is to say that the needs in question need not be merely food or shelter, for example, but also common human needs like self-expression, knowledge, or community.¹⁸ I will assume that necessary productive activity captures a core cluster of necessary features.¹⁹ This characterization is, correctly, more broad than the contours one could recover by considering whatever job descriptions are near at hand; much of our work is unpaid and resists formalization into anything like a job contract with a specified catalog of tasks, but rather happens in the home, public commons, or background of paid jobs. For an example of the latter, consider that completing a college degree is work, and is often, ostensibly, about acquiring skills in order to get a job. This characterization is also more broad than an overly economic sense of work, which emphasizes producing commodifiable outputs.²⁰ But this, again, is as it should be: Homemaking, teaching, and so on count as core cases of work even though they do not characteristically produce commodifiable outputs.

3.1 Inexhaustibly Meaningful Work

Necessary productive activity – work – can be inexhaustibly meaningful. For example, nursing, teaching, carpentry, and philosophy all count as work, and can be sources of meaning for those who do them as long as they continue to do them. This is so even while, in doing such work, workers are often focused on completing particular tasks – such as nursing a patient back to health after a major surgery, or building a piece of furniture, or refuting a philosophical theory.

The practice view makes clear sense of the structure of such inexhaustibly meaningful work, in which workers focus on and complete particular tasks as a way of doing that work. It does so by clarifying that those completable tasks are actionable goals, the upshots of wide specification of the work the worker is doing. Now consider the attitude a person should have toward these goals. These goals are going to be over-specified vis-à-vis the inexhaustibly meaningful work in question, but render its pursuit possible through ordinary planning. It seems appropriate for her to hold these goals at something like an arm’s length. On the one hand, she adopts them to enable her to do her work by taking concrete steps and means; on the other, they should be revised as, in the course of her work, she discovers how they can be replaced by further, superior but still over-specified, goals.²¹

Now, these actionable goals will be completable, and so success can amount to an achievement. For example: when the activity of doing good philosophy is over-specified by the goal of writing a book on Kant’s Groundwork, completing the book can be an achievement. However, there will evidently be a wide range of cases in which a person learns, in the course of her pursuit of an actionable goal, that that goal is not worth completing. So, for example, in trying to concoct a novel counterexample to hedonism, a person might discover that there is little value in continuing to case-monger against ever-more-elaborate variants of hedonism, because disagreements over hedonism do not turn, really, on cases. For her, concocting a novel counterexample – achieving her goal – would not contribute to the meaningfulness of what she is doing; she had better find a new way to continue doing good philosophy. Of course, there will be cases in which the actionable goal does not get overturned in the course of its pursuit, but its completion remains of a piece with the inexhaustibly meaningful work in question. Completing a new interpretation of Kant’s Groundwork, or nursing a patient back to health after a major surgery, or building a sturdy but ergonomic chair can all be the culminations of particular stretches of doing inexhaustibly meaningful work.

But in such cases, it is hard to see how these achievements contribute to the meaningfulness of what one is doing. The order of explanation appears to be rather the other way around: Achieving such goals is meaningful because of the inexhaustibly meaningful work that, in achieving them, one will be doing. It is meaningful to complete a new interpretation of Kant’s Groundwork, or to nurse a patient back to health, and so on, because, in doing so, one is doing good philosophy, caring for another person, and so on.²² This suggests, then, that the practice view better reflects the structure of meaningful work than does achievementism, and thereby offers a more compelling explanation of when and why, when this work takes the form of completable tasks, completing those tasks is meaningful.

Let me briefly handle a pair of related objections. Recall that the appeal to wide specification was meant to recover the possibility of progress in inexhaustibly meaningful activities in general: By being open to, and learning from, experience in pursuing an actionable but over-specified goal, we develop a richer understanding of the standards in light of which further actionable but over-specified goals should be specified going forward. So one way to preserve an explanatory role for achievements might be to say that these lessons constitute achievements. When one learns, for example, how and why case-mongering about wellbeing is methodologically limited (vis-à-vis doing good philosophy), or how and why a particular method of medical care is limited (vis-à-vis being a good nurse), one is achieving something: a richer understanding of the standards for doing good philosophy or being a good nurse. So the thought might go.²³ A related thought is that performing an inexhaustibly meaningful work it itself an achievement: It is an achievement, for example, to do good philosophy, to be a good nurse, or to be an excellent carpenter.

However, neither of these ideas gets much purchase. Recall, after all, that we are widely specifying actionable goals to enable us to pursue inexhaustibly meaningful work, which intuitively has an open-ended character. For example, we may provisionally take up the actionable goal of providing counterexamples to hedonism, because it makes doing good philosophy possible in its limited way. The latter activity is open-ended partly in the sense that it is not completable: One can simply continue to do good philosophy (rather than take up that goal again). But if this is right, then there simply is no way to bring it to a close by completing it. This means, per Sect. 1, that there’s no thing one is supposed to have achieved. Doing good philosophy, or being a good nurse, or excellent carpentry are simply not things we can complete or finish, hence are not things we can achieve.

Importantly, much the same goes for learning from experience. After all, inexhaustibly meaningful work allows for indefinitely refining and expanding the standards of its pursuit, where we learn how this is to go in the course of the work itself. This is the other facet of these activities’ open-endedness. So we cannot finish learning from experience in such a pursuit, but only advance. It therefore is also not something we can achieve. Now, particular lessons can surely be moments of advance, because our understanding has grown deeper. But here, again, saying that these moments of deeper understanding qua achievements contribute to meaningfulness gets the order of explanation the wrong way around. Learning from experience is how we refine our standards in doing inexhaustibly meaningful work, and particular lessons can be achievements of further refinement; but they are themselves meaningful because they are parts of the performance of inexhaustibly meaningful work. They do not themselves explain the inexhaustible meaningfulness of such work.

That is the first round of my argument against achievementism and in favor of the practice view. Inexhaustibly meaningful work is well captured by the practice view’s construal of it as being one or another open-ended practice, which it not something we can achieve. Moreover, concerning the completable tasks we might achieve in the performance of such work, the practice view better explains when and why achieving them would be meaningful: because of the broader, inexhaustibly meaningful work that, in achieving them, one is doing.

3.2 Exhaustibly Meaningful Work

Can there be work that is only an exhaustible source of meaning? Such work might be productive activity that addresses a need that, once met, goes away. The needs I have been emphasizing are not like that, of course – our need for self-expression, knowledge, community, mutual care, etc. are not met and then extinguished. But, of course, some of our needs are plausibly extinguished once met – for example, the need for more effective treatments for cancer. Work that addresses such needs might be meaningful only while such needs are unmet.

It is, however, a bit difficult to say when, and why, such work could be only exhaustibly meaningful, simply because addressing these sorts of needs is standardly of a piece with broader, inexhaustibly meaningful work. After all, searching for more effective cancer treatments is of a piece with caring for one another, and with scientific inquiry in general.

So let me build up a positive proposal. Many activities are thoroughly shared, where each individual only does their part in that shared activity. To illustrate with a well-known example, our dancing the tango is a shared activity; what you and I each do is our part in it, and, strictly speaking, neither of us, individually, dances the tango. As the saying goes, it takes two. Similarly, inexhaustibly meaningful work like scientific inquiry, mutual care, and so on are shared activities. So, strictly speaking, there may be cases in which we do not want to say, of any particular scientist or care-worker, that she is doing scientific inquiry or mutual care; rather, we want to say she is doing her part in shared scientific inquiry or mutual care. This cleaves the inexhaustibly meaningful shared activity from an individual’s part in it. It would then not follow, from the fact that a person is doing her part in an inexhaustibly meaningful shared activity, that her individual work is inexhaustibly meaningful.

Nonetheless, it would be very surprising if doing our part in inexhaustibly meaningful shared activities was never, itself, inexhaustibly meaningful. But what would make our participation inexhaustibly meaningful? Recall that, in inexhaustibly meaningful activity, we widely specify actionable goals to pursue, because pursuing those makes pursuing the activity possible. Now, in the case of inexhaustibly meaningful shared activities, this wide specification will itself often be shared.²⁴ I suggest that, when any individual participant is participating in such shared wide specification, this suffices for her pursuit of the downstream, actionable goals to be inexhaustibly meaningful. This is because these actionable goals will be shaped, in significant ways, by her own practical point of view, as are their superior replacements that come later on the basis of learning by experience.

If this suggestion is on the right track, we can now provide a model for work that is only exhaustibly meaningful. Return to the cancer-treatment example, and recall that the scientist’s search for a more effective cancer treatment will not continue to be meaningful after, and if, such a treatment is discovered. And this is because, by discovering such a treatment, she will meet a need that then goes away.²⁵ What would render her work only exhaustibly meaningful, on the model I am offering, is the further supposition that she is not a participant in the shared wide specification of that very goal. When that goal (along with concomitant standards) is handed down to her in actionable form, her pursuit of it (in accord with those standards) will not be inexhaustibly meaningful. And this is so even if, as is natural, we want to say that her pursuit of it is of a piece with a broader shared activity that is inexhaustibly meaningful work.

So we have before us a model of exhaustibly meaningful work, which requires the proviso that the worker in question does not participate in the shared wide specification of the goal that structures her work. Very briefly, permit me a few stray remarks that I will pick up again in due time. Notice that, in principle, the pursuit of such a goal can be fully automated. This is because the goal is suitably specified, along with the standards for the pursuit of it, in such a way that it appears amenable to proceduralization and automatic execution. And notice that such a worker has already been displaced²⁶ – not by automating technology, but rather by whoever is calling the shots. Lastly, notice that her work is beginning to look very much like a job.

What, if anything, does achievement contribute to the exhaustible meaningfulness of such work? Now, the goal in question – which I have so far described only as ‘a more effective cancer treatment’ – needs much more specification to even get the case off the ground. After all, ‘effective’ invokes a standard itself in need of specification, which can proceed in terms of relative fatality rates, relative remission rates, painfulness and discomfort during treatment, and even broader concerns like affordability, feasibility, the likelihood of the patient’s full cooperation, and so forth. Moreover, ‘cancer’ is itself an umbrella term for various illnesses with various sources, risks, and treatment possibilities; thus, the goal also needs to be specified with respect to what kinds of cancers such treatment is being sought for, and which treatment modalities are to be investigated, and how (see Bilby 2019). As I have been emphasizing, this wide specification proceeds in the course of doing inexhaustibly meaningful work, and, I add, is itself an ongoing process that is reflected in the thickness of the evaluative concepts in play.

The subtleties of wide specification are going to crop up, now and again, in what follows. But let us deal with the simplest case first. Considering a worker pursuing a goal that has already been given the requisite specification, her goal is always subject to obsolescence or modification in favor of superior (but unavoidably over-specified) actionable goals. When novelty undermines the quality of the worker’s goal, which here means that it no longer suitably serves the inexhaustibly meaningful work for which she is doing her part, it should be clear that her achievement of that goal does not matter. As far as meaningfulness goes, her contract ought to be renegotiated.

Now, we might wish to suppose that the worker is given some actionable goal that will not be overturned or substantially modified. It’s worth bearing in mind that this supposition is going to be artificial in important respects: Because wide specification of actionable goals relies on our capacity to learn from experience, we do not have good reason to suppose, of any particular such goal, that it will not need to be overturned or substantially modified in light of things we have yet to learn from experience.²⁷ We might learn of some genuinely novel cancer-treatment modality in the course of investigating some more familiar one, for example, where the former holds out promise of being more effective in some relevant sense.

Nevertheless, let us forge ahead by supposing that novelty will not undermine the quality of her goal or warrant significant alterations thereto. We could then ask whether there is anything relevant to be learned from experience in the pursuit of that goal. However, because we are considering a case of exhaustibly meaningful work, which we constructed by supposing that she does not participate in the wide specification that determined her job, we do not have an uptake mechanism in place for such lessons. Or, more carefully: Whatever the worker learns from experience in pursuing the goal will not guide the wide specification of further goals for the inexhaustibly meaningful work that (at some remove) structures and motivates her work in the first place. So, in our present example, this means that whatever is learned about how to conduct scientific inquiry, or how to care for one another, it will only come by observation of the worker, or by examination of her outputs – not by her own contributions on the basis of learning from experience. Whatever lessons the worker herself might learn, the organization of her workplace divorces them from the broader (shared) activity in virtue of which her work is meaningful; thus, however much such lessons are themselves achievements, this fact does not promise to explain anything about the exhaustible meaningfulness of her work.²⁸

Might her achievement of the task given to her nonetheless explain why her work is exhaustibly meaningful? Perhaps, but it would depend on the details. Adina Schwartz has argued that workplace arrangements in which workers’ tasks are rigidly defined by others – as on an assembly line, for example – deprive those workers of meaningful work, because they are not able to exercise their autonomy in performing their work and delineating its character (Schwartz 1982). The example we are currently discussing perhaps allows for autonomy in the performance of the job: Although the worker has been tasked with a suitable specification of the goal of finding a more effective cancer treatment, she may have considerable leeway in working out the means and methods whereby she will do this. For the sake of argument, suppose this is so. Accordingly, achieving her task might be meaningful. But we have an explanation as to why it would be: the simple fact that, in succeeding, she is doing her part to advance scientific inquiry and to care for one another, the inexhaustibly meaningful shared activities in the background that (at some remove) give her job its structure in the first place. The meaningfulness of her success is explained by the broader, inexhaustibly meaningful shared activity she thereby participates in; her success does not explain, hence does not contribute to, the meaningfulness of what she is doing. Achievementism again gets the explanation the wrong way around.

Suppose, on the other hand, that this worker does not have enough leeway in her job to exercise her own autonomy – namely, that her job is so thoroughly specified that performing it is a matter of following a set procedure. In other words, suppose that her job is amenable to being fully automated away. Now, if the automating technology does not exist, and if the task is sufficiently important – as developing more effective cancer treatments clearly is – we might want to say that her performance of it is meaningful, her lack of autonomy notwithstanding. The first of these conditions bears considerable load: If the automating technology does exist, why not automate her tasks? It is rather difficult to see how her work could be meaningful when such tasks remain hers to perform. As far as meaningfulness goes, it is hard to see why we should not wish that she were freed from these tasks to do different, more clearly meaningful, work – or to do something meaningful that isn’t work at all.²⁹

This case presages what Danaher and Nyholm call ‘total replacement’ of a worker by an automating technology (2021: 229). According to them, total replacement threatens meaningful work, because “[w]hatever achievements [the totally replaced worker] used to associate with their job are now closed to them” (2021: 232). Under achievementism, the argument is clear: All else equal, total replacement cuts a worker off from achievements in her work, and so from something that would otherwise contribute to the meaningfulness of that work. The problem is that the present case presses against achievementism itself. We are, after all, dealing with a case in which a worker performs a job that has been fully specified for her in advance, its means, methods, and goal so thoroughly articulated that it could, in principle, be fully automated. Moreover, whatever lessons she is to learn in the course of performing this job she will be keeping to herself, because she is not a participant in the wide specification that created her job in the first place. The saving grace of her job appears to just be that we do not yet have the technology to automate it. Thus, the meaningfulness of her job appears already intensely precarious, and easily undermined by disruptions in the worthwhileness of that job or the technology on hand. Achievementism might appear at first blush to get something right about this, because it allows us to say that such a worker at least has the opportunity to achieve in her job. Importantly, achievementism has this implication even when the automating technology is on hand, but simply not in use. And that means that there is at least some reason not to use such technology, so that the worker might still achieve in her job. This should strike us as strange.

One final strategy for defending achievementism here is to insist that other constituents of achievement – the challenging, substantially and non-lucky causing of the relevant outcomes – are explanatorily relevant. There are two things to say in reply. First, bear in mind the construction of the case. We are supposing that her goal has already been widely specified and simply given to her, and that it will turn out that nothing learned from experience in the pursuit of that goal will shape downstream wide specification. Neither supposition, as I have been emphasizing, comports well with how meaningful work unfolds in the real world. So we can already say: If this is the last refuge for achievement to play an explanatory role, it does not amount to much. Second, and since we are already supposing that lessons learned from experience will get no uptake, what is the relevance of such challenge and skill to the inexhaustibly meaningful work in the background that structures the job? It cannot be that, in meeting whatever challenges arise, or in deploying her skills competently, the worker will discover something important about how that inexhaustibly meaningful work is to be done. To be sure, overcoming challenges and competently deploying her skills might be means to success in her job, but this will not be enough. We are looking for an explanation of what makes her work especially meaningful. We are not looking for a description of the instruments with which she does especially meaningful work. So if this is all their relevance comes to, then they do not support the idea that achievements make work especially meaningful.

That is the second round of my argument against using achievementism to assess meaningful work. After constructing a model of merely exhaustibly meaningful work as a kind of subordinate participation in broader, shared, inexhaustibly meaningful work, I argued that achievementism gets a grip in only a certain kind of case. In particular, it gets a grip when a worker’s tasks are simply given to her in fully actionable form by other participants, and these tasks turn out not to be overturned at any stage prior to their completion. However, we are then faced with a task that can, in principle, be automated away. In the absence of such automating technology, one might insist that achieving such tasks can still be meaningful. So the final turn of my argument was that, even if this is so, the achievement does not explain why completing the task is especially meaningful; rather, it highlights how a worker completes the task.

4 Conclusion

Achievementism should not be used to assess meaningful work after automation. This is for two reasons. The first is that a wide range of meaningful work cannot, properly speaking, be achieved at all. This is because the work cannot be completed. The second is that, for meaningful work that can be completed, achievementism gets the explanation the wrong way around. In these cases, what explains why completing the work is meaningful is the broader, non-completable meaningful work of which it is a part. The practice view, by contrast, does not suffer these flaws. It puts non-completable meaningful work front and center, and better captures how that work can structure downstream, completable meaningful work.

Let me conclude with one final virtue of the practice view of meaningful work. We can now appreciate why some meaningful work is amenable to automation, in much the same terms as why it is only exhaustibly meaningful. When completable tasks can be handed down to a worker by other participants of some meaningful shared activity, where her own practical thinking and capacities for learning from experience need not guide the completion of that task or further wide specification of downstream tasks, those tasks can be schematized in the manner of an algorithmic procedure. What is to be accomplished, and how it is to be accomplished, is simply given. This means that automation is an endemic threat to exhaustibly meaningful work.

This is a substantial reconceptualization of the unease over the future of meaningful work in the face of rapid advancements in automation. Unlike the initial worry with which this paper began, this concern does not turn on questionable empirical presuppositions about what automation will do to employment in the future; nor does it turn on a misguided commitment to the especial meaningfulness of workplace achievements. Rather, it turns on how we work together to meet our needs – for food, shelter, understanding, community, whatever.

Moreover, despite the endemic threat that automation poses to such forms of meaningful work, the ethical action is largely elsewhere. In cases where a worker’s job is amenable to automation, her position as someone doing meaningful work is already intensely precarious. To the extent that we want to expand opportunities for meaningful work, as we should, it is better to improve her precarious position than resist automating her job.

By contrast, inexhaustibly meaningful work exhibits open-endedness – in particular, the standards are worked out in the work itself – and so is not at risk of automation itself. This is because a procedure can only be automated as long as the steps are defined, whereas, in open-ended activity, which steps to take is indefinitely open to redefinition. Thus, one promising way to improve workers’ position is to reconfigure work in a way that strengthens its connection to inexhaustible meaningfulness. As described in Sect. 3.2, this would be a matter of empowering workers to participate in the wide specification of the actionable goals we pursue in doing inexhaustibly meaningful shared activities.³⁰ This is a way of ensuring workers maintain a significant degree of creative control over their own work; in the event that the completion of well-defined tasks does indeed get outsourced to automation, such workers would then stand in a creative, supervisory role vis-à-vis the machines. However, since, as I argued in Sect. 3.1, inexhaustibly meaningful work does not depend on achievements, machines’ completing such tasks does not threaten the meaningfulness of what the human creative directors do.

Returning to the distinction between work and a job, moreover, we should bear in mind that, as automating technologies advance and render various jobs no longer necessarily performed by persons at all, many such former workers would simply opt to do meaningful work that is ill-suited to employment with a job contract in the first place – such as homemaking, or caretaking – or to do meaningful activities that are not work at all – such as art-making, or gameplaying. Measures, such as a universal basic income, ameliorate a rather different (economic) precarity that otherwise forecloses on such especially meaningful activities. There are, of course, many other arguments in favor of such measures. The focus here is just on the aim of supporting especially meaningful activity.

Supporting especially meaningful activity is not a matter of casting a critical eye on emerging automating technologies in the workplace, but rather on the precarity of those workers whose work is amenable to automation. For those privileged enough to already be engaged in inexhaustibly meaningful work, automation is not, in principle, a threat. However, and of course, automation may pose an indirect threat to them as well when political-economic forces press for the total replacement of such workers with automated technology – for the replacement of nurses with ‘care’ robots, for example. But the issue here, as I have argued, is not that ‘care’ robots will do the very same work that would otherwise be done by nurses. Rather, the issue is that ‘care’ robots do not really care at all, and so the inexhaustibly meaningful activity would simply cease to be done.