Freedom and the Machine: Technological Criticisms in Adam Smith’s Thought

The Division of Labor and Technology

Smith’s arguments in his Theory of Moral Sentiments and The Wealth of Nations offer many critiques of an unchecked commercial system. Donald Winch argues that Smith’s complex account of the emergence of commercial society includes the explicit caveat that “the new form of society is marked by several major defects of a moral and civic character”(Winch 1978). Though Smith’s critiques of the negative effects of the division of labor, primarily the mental mutilation of the working class, are well-known (Smith 2010; Herzog 2016; Hurtado 2019), there is comparatively little discussion in the Smith literature on a particular component or product of the division of labor: technology itself in the form of productive machines. Smith argues that technological advancements are largely a natural consequence of the division of labor; when people are locked into a small task, they tend to innovate, creating timesaving, production-increasing devices (WN I.i.8). This innovation further drives the specialization inherent in the division of labor, allowing each person to take on successively smaller tasks. Technology is then not simply one piece of the overall puzzle of commercial society but is rather a central component of the advancement and growth of production. Because of this, Smith’s criticisms of the consequences of the division of labor must be taken in concert with the advancement of technology; the division of labor and resulting technology are so closely related, and technology itself so important in the narrative of the progress of production, that we ought to consider the role of productive technology more explicitly and precisely in Smith’s thought. Smith’s description of innovation by common workers further suggests that the division of labor becomes significantly mentally mutilating on his account only after extensive technological advancement.

Smith argues that increased production following the division of labor into smaller components is “owing to three different circumstances; first, to the increase in dexterity in every particular workman; second, to the saving of the time which is commonly lost in passing from one species of work to another; and lastly, to the invention of a great number of machines which facilitate and abridge labor, and enable one man to do the work of many” (WN I.i.5). The first two points are intuitive. Smith’s third point is both that machines increase the productivity of labor in society where labor is divided, and that the invention of these machines is encouraged by the division of labor itself. That machines increase production is likewise so commonsensical that Smith claims, “It is unnecessary to give any example” (WN I.i.8). He chooses instead to reflect on the fact that “the invention of all those machines by which labor is so much facilitated and abridged, seems to have been originally owing to the division of labor” (WN I.i.8). Smith reasons that a worker’s focused attention to a specific task will almost inevitably lead to the invention of various means to perform that single task more efficiently.

Smith’s examination is primarily descriptive but somewhat laudatory. The division of labor is the cause of “the greatest improvement in the productive powers of labour” (WN I.i.1). This is a net positive in many respects. Those “civilized and thriving nations” (WN “Introduction and Plan of the Work,” paragraph 4) where people thrive rather than subsist are of course in some way preferable to countries where many of the poorest classes are languishing for want of the benefits the division of labor offers. However, Smith devotes a significant section of the latter part of WN to exploring the negative effects of the division of labor that must be counteracted in some way, as he is “keenly aware of the potential for technical advancement in production to numb the mind of the workers” (Evensky 2005, 13).

Smith again states the obvious: when labor becomes divided in commercial society, “the employment of the far greater part of those who live by labour, that is, of the great body of the people, comes to be confined to a few very simple operations; frequently one or two” (WN V.i.f.50). One person in manufacturing might spend the bulk of his work hours operating a lever or joining two things together repeatedly. Although this repetition does make the worker dexterous, it also has the side effect of being inherently boring. Smith notes that when labor has been divided to this extent, the worker “has no occasion to exert his understanding, or to exercise his invention in finding out expedients for removing difficulties which never occur” (WN V.i.f.50). Either a worker or the machine itself performs a small task, guided by a human overseer, with no room for creativity or ingenuity. All inefficiency, which itself occasions ingenious innovation, has been or will be purged by design.

The consequences of this extreme specialization are drastic. Though this worker might make enough wages to survive, his life is pitiful at best. Smith describes a man who is

. . . not only incapable of relishing or bearing a part in any rational conversation, but of conceiving of any generous, noble, or tender sentiment, and consequently of forming any just judgment concerning many even of the ordinary duties of private life. . . His dexterity at his own particular trade seems, in this manner, to be acquired at the expence of his intellectual, social, and martial virtues” (WN V.i.f.50).

Though there may be standouts in any society, “Notwithstanding the great abilities of those few, all the nobler parts of the human character may be, in a great measure, obliterated and extinguished in the great body of people” (WN V.i.f.51). Smith here explicitly invokes judgment, saying that a man in such a situation cannot properly judge private or public interests (WN V.i.f.50). That is, the economically beneficial advances in technology have caused harms to the ability of the common person to refine and exercise judgment.

It is thus advanced commercial society, and the technology that can only come about in such a society, that experiences the harms caused by over-specialization. In previous stages of economic growth, such as the “barbarous societies. . . of hunters, of shepherds, and even of husbandmen. . . the varied occupations of every man oblige every man to exert his capacity, and to invent expedients for removing difficulties constantly occurring” in a way that occupations in commercial society do not (WN V.i.f.51). Despite material poverty compared to the wealth of commercial society, so-called “barbarous” societies have the advantage of engaging workers’ minds constantly, preventing them from becoming inhuman through repetition and boredom.

Familiarity with the well-known story of productive advancement that Smith tells can belie the dependence of the extended division of labor on machinery. Without efficient ways of extracting ore, smelting useful metal, and transporting that metal, a blacksmith could hardly be put to the work of making nothing but nails day in and day out, which naturally precedes the division of the task of nail-making into even smaller tasks. These technological preconditions are exactly what Smith describes in his early draft of the Wealth of Nations (ED, 568). In a similar vein, Smith argues “That the Division of Labour is limited by the Extent of the Market” (WN I.iii), and that the extent of the market is in large part determined by access to goods. This access to goods is, in turn, technologically dependent, for it is “by means of water-carriage” that “a more extensive market is opened to every sort of industry than what land-carriage alone can afford it” (WN I.iii.3). It is the improvement of technology that allows for labor to subdivide to the great extent that Smith fears, and it is the later technological progress, occasioned by the earlier, that leads to the over-specialization that seems to be most concerning for Smith. ⁴

Thus, though the stupefying effects Smith describes are unique to commercial society, the ability to subdivide labor to the extent that Smith finds problematic is distinctively the result of technological advancement. At the same time, Smith argues that the division of labor occasions an engagement of the mind that inspires innovation. In fact, the development of the greater part of those “machines” comes from, he argues, those who have been put to very onerous or demanding or boring tasks. This is seen in the example of the small boy automating his task out of boredom and a desire to play (WN I.i.8). The boy, already engaged in machine work, has the occasion to exercise his ingenuity and improve upon the machines already at hand. This is seen, too, in the example from the early draft of the Wealth of Nations of the wretched slave inventing the hand-mill. It was probably, Smith says, “Some miserable slave, condemned to grind corn between to stones by meer strength of his arms. . . who first thought of. . . the original, rude form of hand-mills” (ED, 569). In both cases, those enjoying both the benefits and the inconveniences of rudimentary machines employ their understanding to create more complex and effective machines. In other words, machine technology both precedes and is occasioned by significant division of labor in Smith’s account.

In his description of the development of machines, then, Smith shows that at a certain stage, creative ingenuity, or what Smith calls philosophical activity on the part of the worker (ED, 570), continues despite the division of labor. A task that is not yet as efficient as it could be may yet be improved by the invention and application of machinery. People like the creative boy automating his job are still encountering challenges that need resolution. As shortcuts and improvements are made in production, subsequent workers and masters both benefit from the increased productive capabilities enabled by these inventions. At a certain point in Smith’s story, however, the division of labor occasions the development of so many machines or such meticulous machines that workers are forced into narrow tasks that can no longer be improved. Workers then become less and less human in proportion to the advances and simplifications introduced in production, absent some sort of external intervention.

This stage, I argue, is where mental mutilation becomes a serious and widespread concern for Smith. At previous stages of production, the occasions for exerting mental effort in various directions arise naturally and regularly. When machines proliferate, those opportunities are fewer and farther between, and require far more skill and ingenuity to develop. It is this extensive technological proliferation that then locks a worker into an over-specialized and mentally mutilating task. At first glance, it may seem that Smith’s analysis of the division of labor is primarily concerned with the degradation of the workers' practical capabilities, what Ron Westrum calls "deskilling" in his analysis of Marx's reception of Smith (Westrum 1991, 33). But Smith seems unconcerned with the mere fact of possessing a skill, but rather with what the specialization of labor due to technological progress is going to do to the worker's mind and character. The solution of public education to stave off mental mutilation and preserve “the proper use of the intellectual faculties of a man” and the “essential part of the character of human nature” (WN V.i.f.61) demonstrates this. Certainly, some of Smith’s concerns are practical; a society in such a degraded position with swaths of over-specialized citizens would be in constant danger should anything out of the ordinary, such as sudden war or economic disruption, place strenuous demands on the working class. However, even if such a practical external difficulty never arose, a society filled with uninteresting people with none of the “nobler parts of human character” would simply be a morally poor society. As Smith argues, “No society can surely be flourishing and happy, of which the far greater part of the members are poor and miserable” (WN I.viii.36). In light of Smith’s concern for the worker’s mental state and humanity, surely no society can be flourishing and happy where the workers, sufficiently materially supplied, are miserable due to mental mutilation. Smith’s thought then exposes how increased reliance on machines harms the individual’s ability to sympathize with fellow citizens, judge independently, and achieve happiness in a free commercial society. ⁵ Scholars such as Dennis Rasmussen have highlighted this: “There are actually two separate problems here, for Smith: the worry that a refined civilized populace might be unable to defend itself against more rugged, barbarous peoples, and the concern about the deformity this feebleness causes in people’s characters” (Rasmussen 2005, 112). On Rasmussen’s account, Smith’s concern is with the dignity and equality of the workers in this new commercial society. But for Smith, there is an intrinsic connection between “the proper use of the intellectual faculties of a man,” the “intellectual, social, and martial virtues,” and freedom in Smith’s account of the development of sympathy and judgment.

Imaginative Sympathy and Judgment

In TMS, Smith argues that the faculty of sympathy is what enables people to be interested in “the fortune of others” (TMS I.i.1.1). Human potential, on Smith’s account, is best fulfilled through the progressive development of faculties for sympathy, moral reasoning, and judgment, which allow people to associate freely and experience the joys of human sociability. This natural ability to feel with or for other people in both their happiness and their sadness, their pleasure and their pain is what he calls sympathy. The subsequent arguments in TMS provide detailed accounts of how precisely people learn to moderate their passions sympathetically so as not to overwhelm spectators with joy or grief in excess, beyond the spectator’s ability to sympathize (TMS I.i.4.8). Smith further argues, “that to feel much for others and little for ourselves, that to restrain our selfish, and to indulge our benevolent affections, constitutes the perfection of human nature; and can alone produce among mankind the harmony of sentiments and passions in which consists their whole grace and propriety” (TMS I.i.5.5). Because people desire to have compassionate compatriots, they will oblige others by restraining their own emotions when tempted to be excessive and take care to notice the emotions of others. One who has fully developed the virtues of care for others and self-command has achieved, in Smith’s words, the perfection of human nature.

In this account of sympathy leading to the development of virtue, we see the importance of judgment for Smith’s project. The role of judgment in Smith’s thought and the importance of the exercise of judgment for meaningful freedom is explored at length by Samuel Fleischacker in his book A Third Concept of Liberty. ⁶ Fleischacker writes that though judgment often evades precise definition, a working definition of judgment is “the conclusion of a train of thought where the interpretation of particular cases is essential to that train of thought, and the practice under whose rules the conclusion is reached therefore allows for further reasonable disagreement over that conclusion”(Fleischacker 1999, 9). Fleischacker unpacks notions of judgment among Smith’s predecessors and contemporaries in order to demonstrate that judgment “is informed by perception and reason but identical with neither” (Fleischacker 1999, 127). The process of imaginative, sympathetic apprehension is then merely the starting point, and the choices made on the basis of sympathy constitute the exercise of judgment. In the act of choosing between a number of complexes, particular options after apprehending and reasoning, the individual exercises judgment. Judgment on this account cannot be unreflective or random because it requires interpretation of a particular situation and subsequent intentional action. It is not purely algorithmic or calculative both because it rests strongly on the sentiments and because there can and likely will be reasonable disagreement over the judgment made. For Smith, one becomes a better human being by exposing oneself to the criticisms and corrections of spectators, real and imagined, in our emotional, social, and economic lives.

Fleischacker writes that this “third” concept of liberty “construes freedom above all as that which enables one to judge for oneself—unlike a child, who requires others to judge for her, who requires tutelage” (Fleischacker 1999, 9). The alternative to the exercise of independent judgment is a kind of bondage, either to the judgments of another person or thing or to the whims of appetite; in other words, dependence. Importantly, judgment of the sort that makes one meaningfully free is not precisely limited to one sphere or type of judgment in Smith’s account. We see Smith’s thoughts on economic judgment in his arguments in favor of freedom for workers to choose their professions. He argues, for example, that restrictions on when and where someone can apply for employment are “a manifest encroachment upon the just liberty of both the workman, and of those who might be disposed to employ him,” because it restricts the exercise of judgment on the part of both the worker and the employer (WN I.x.c.12). Neither the worker nor the potential employer is fully free to judge for themselves, or to submit their judgment to the criticisms of others. The use of judgment described in this section of WN mirrors the cooperative feedback mechanism of moral development in TMS in key ways: the prospective worker must submit his own judgment about his own talents to the criticism of spectators, potential employers, whose feedback can then be used to revise the worker’s own judgment in subsequent decision. This focus on the choice of the worker continues when Smith turns to laws restricting the movement of the worker without permission from the government. Smith writes, “To remove a man who has committed no misdemeanor from the parish where he chuses to reside, is an evident violation of natural liberty and justice” (WN I.x.c.59).

Though the relationship between independent judgment and freedom is clear in economic matters, Smith’s arguments extend to moral judgment in TMS as well. Smith describes the development of the virtue of self-command at length, concluding that without the sense of propriety which the “regard to the sentiments of the supposed impartial spectator” provides,

. . .[E]very passion would, upon most occasions, rush headlong. . . to its own gratifications. Anger would follow the suggestions of its own fury; fear those of its own violent agitations. Regard to no time or place would induce vanity to refrain from the loudest and most impertinent ostentation; or voluptuousness from the most open, indecent, and scandalous indulgence (TMS VI. concl.2)

The picture Smith paints is drastic; absent the self-command that restrains the passions, the individual is subject to violent changes of emotion that drive action apart from reflective judgment.

Smith makes clear, then, that the free and independent exercise of judgment is not possible simply through the absence of constraint. True freedom requires instead a process of education and reflection by which people learn to exercise their judgment independently. ⁷ Though I have emphasized so far, along with Fleischacker, that this process aims at independence, it is important to note here that this process of refinement can only be properly or fully undertaken in social situations where the reactions of other people can be observed and their opinions taken into account. Smith explains, “Society and conversation, therefore, are the most powerful remedies for restoring the mind to its tranquility, if, at any time, it has unfortunately lost it; as well as the best preservatives of that equal and happy temper, which is so necessary to self-satisfaction and enjoyment” (TMS I.i.4.10). For though Smith supposes that people can develop internal mechanisms of self-command, personified in the impartial spectator, this requires both an initial social exposure and also a continued one. Ryan Hanley argues that “the moral autonomy that Smith seeks calls only for the transcendence of the opinions of others, and not the transcendence of the active duties to others which he deems the proper responsibility of the virtuous individual”(Hanley 2008, 138). But autonomy, or self-legislation, does not seem to fully capture what Smith is advocating. Though the goal of the refinement of judgment and sharpening of sympathy seems to be a certain kind of independence, achieving that independence requires an ongoing and complex interdependence on other people with reciprocal, social duties. Smith argues, for example, that:

In solitude, we are apt to feel too strongly whatever relates to ourselves. . . The conversation of a friend brings us to a better, that of a stranger to a still better temper. . . the abstract and ideal spectator of our sentiments and conduct, requires often to be awakened and put in mind of his duty, by the presence of a real spectator: and it is always from that spectator, from whom we can expect the least sympathy and indulgence, that we are likely to learn the most complete lesson of self-command (TMS III.3.38).

In other words, self-command on Smith’s account is fundamentally relational, and never fully autonomous. ⁸

Meaningful freedom then requires a complex interaction of interdependent, social individuals who refine their virtues and their judgment with reference to others. No one else is making these judgements on behalf of the individual, but no individual is exercising his judgment fully without reference to his fellows. For Fleischacker, we risk harming the individual’s ability to exercise independent judgment most often when we enact paternalistic government policies that impede or prohibit the developments and the exercise of judgment. Smith’s account of the progress of the market and the development of machine technology would suggest in addition that paternalistic decision-makers can come in the form of complex technologies that first rob the worker of the ability to exercise judgment and then could, in the present day, make substitute decisions in the worker’s stead. If a novel threat to freedom is introduced by the existence of machines that substitute for human ingenuity and narrow the scope of human decision-making, how much more danger is faced when the machines themselves are substituting decisions for those of the worker? In other words, what if anything can Smith’s concerns about new productive machines tell us about our algorithmic AI and machine learning technologies today?

Technology Problematized

I have argued that one significant threat to liberty on Smith’s account is the tendency of new machine technology to harm the individual’s ability to develop and exercise independent judgment. Smith is not alone in identifying technology as a possible threat to liberty, nor is he the only source that might provide resources for contemporary technological critiques from a liberal democratic perspective. For example, Elliot D. Cohen has identified technological threats to liberal freedoms in his study Technology of Oppression; Cohen specifically identifies new surveillance and machine learning technologies employed by as invasive threats to freedom and privacy with “potentially oppressive consequences for the masses” (Cohen 2014, 107). Shoshana Zuboff’s popular work on “surveillance capitalism” addresses many of the same concerns, with an eye towards threats to freedom posed by corporations and their broad surveillance networks (Zuboff 2019). In both cases, the obvious threat to liberty ultimately comes from harmful or malicious use of technology to directly oppress. However, Smith’s arguments, both in regards to over-specialization and in regards to the narrowing of available choices, allow for a more subtle and potentially troubling critique of technology. If Smith’s understanding of freedom is correct, and there are technologies that deprive individuals of their ability to develop or exercise independent judgment, then a threat to liberty need not take the form of overt oppression but could instead be an infectious and pernicious side effect of the use of these technologies.

There are then two ways, based on the foregoing arguments, that one could conceive of machines as a threat to liberty on Smith’s account of freedom: the first would be through causing over-specialization in the working class, depriving them of essential features of humanity. The second would be in substituting external decision-making, the decisions and judgments of others, for individual judgment and thereby in impeding the feedback mechanisms necessary for developing judgment by removing occasions for people to submit their decisions to social approval. It is the former harm that Smith explicitly connects to machines; it is the latter that our algorithmic technologies risk perpetrating.

To be clear, though some today might take part in over-specialized labor as Smith described, this effect need not necessarily come from the over-specialization that Smith feared and that the technologies he was familiar with tended to encourage. ⁹ And importantly, Smith’s arguments do not suggest that all tool-using or dependence on technology commits the same meaningful harms as these. The primary threat would come from a narrowing of the scope of the exercise of judgment by substituting external judgments for those of the individual in increasing numbers, and moreover a training of the individual not just to augment their productive ability but to depend in a substantial way on those judgments of others, shortcutting the developmental process that Smith describes. Thus, if machines today are substituting external decision-making for individual judgment, preventing opportunities to develop judgment as extensively as might have been previously possible, those who use and are influenced by such machines would be at risk of losing their liberty.

If Smith was concerned with the direct degradation of the worker's ability to exercise judgment, he was also concerned with preserving the worker's opportunity to exercise judgment in his daily life. In fact, the two are codependent; the ability to judge independently is premised on available space to make these judgments, submitted to external criticism. On Smith’s notion of a system of natural liberty, individuals ought to be afforded the opportunity to make decisions for themselves based on their own judgment, which is presumably better informed than that of the bureaucrats. As Smith puts it, “. . . the law ought always to trust people with the care of their own interest, as in their local situations, they must generally be able to judge better of it than the legislator can do” (WN IV.v.b.16). This daily exercise of judgment is what is in danger of reduction or elimination due to the automatization that is exclusively allowed by contemporary technology.

An argument to this effect has recently been proffered (Helbing et al. 2019). As researchers on the cutting edge of their respective fields, Helbing et al. argue that contemporary algorithmic technologies, which are increasingly used to make choices in matters from navigation to medical treatment, risk imposing immaturity, or a deprivation of independent judgment, on swaths of unsuspecting consumers and specialists alike. Helbing et al. highlight the extreme and extensive influence automated processes, driven by AI and machine learning, have on individual choice-making in a section suggestively entitled “Programmed Society, Programmed citizens.” The application of artificial intelligence and machine learning consumers are most likely to encounter on a daily basis is predictive and suggestive. Online shopping and entertainment options track user behavior and progressively learn what the user prefers. This information is then used to suggest new products and new entertainment options. In some cases, algorithms seem to know us “possibly even better than our friends and family or even ourselves” (Helbing et al. 2019, 75). The obvious problem here is one of free and informed choice-making. ¹⁰ The writers explain, “Often the recommendations we are offered fit so well that the resulting decisions feel as if they were our own, even though they are actually not our decisions. In fact, we are being remotely controlled ever more successfully in this manner. The more is known about us, the less likely our choices are to be free and not predetermined by others” (Helbing et al. 2019, 75). ¹¹

Though many forms of targeting are based on prior choices, there is a fine line between suggestion and persuasion, or persuasion and pressure, or pressure and outright manipulation. In fact, Helbing et al. immediately note that technology companies are moving ever closer to large-scale persuasive efforts rather than purely predictive algorithms. They write, “In the future, using sophisticated manipulation technologies, these platforms will be able to steer us through entire courses of action. . . The trend goes from programming computers to programming people” (Helbing et al. 2019, 75, 76). This type of widespread, subtle psychological manipulation would certainly be troubling for Smith. It prevents people from making fully informed judgments about their own economic, moral, and political lives, even if they are not in a type of employment that specifically mentally mutilates them. The mental mutilation instead comes from another source: pre-programmed mechanisms that make choices on behalf of the individual, while lulling the individual into a false sense of personal autonomy. These systems make the choices of supposedly free individuals dependent on and determined by outside forces that cannot be directly apprehended, reasoned about, or easily criticized by laypeople.

Even worse, on Smith’s account, the choices made by these machines are not made socially in the same way human judgments would be and are thus bereft of necessary interdependent feedback based on sentiment and virtue rather than mere calculation. Returning to the process of developing judgment outlined previously, Smith argues at length that our judgments must be subject to external critique in order for the process of honing judgment outlined above to be successful. As Fleischacker argues by extension, this is what makes judgment distinct from pure algorithmic reason (Fleischacker 1999, 9). Individuals can then modify their behavior and their passions in accordance with reactions of spectators, and can only do this fully and well in the presence of real spectators whose opinions, both positive and negative, are taken into account in future judgments. This system of external critique is what allows for the refinement of judgment. But the necessarily opaque and impenetrable nature of algorithmic decision-making introduces clear problems, depriving us of the feedback and criticism necessary to produce good judgment.

This has implications both for machine learning itself, where a program “learns” self-referentially without external criticism, and for those who employ these technologies to make significant decisions. As Matthew B. Crawford argues, “human beings are able to give an account of their reasoning. . . Judgments made by an algorithm (ideally one supplied by a third-party vendor) are ones that nobody has to take responsibility for” (Crawford 2019). When one is conditioned to make decisions on the basis of the output of an algorithm, there is a risk of blame-shifting and an absence of accountability; this is a phenomenon that Rubel et al. have termed “agency laundering,” where the agency of the decisionmaker is “laundered” and attributed to the ledger of the machine to cleanse the agent of moral responsibility. The judge who makes a ruling based on predicted probabilities of recidivism calculated by an algorithm, or the doctor who prescribes medication with the same method need not be held to account for that choice but can instead address any resultant challenge to a different decisional agent: the machine. The risk is that “Using an algorithm to make decisions can allow an agent to distance herself from morally suspect actions by attributing morally relevant characteristics to the algorithm” (Rubel, Castro, and Pham 2020, 564). The broader concerns at play in an algorithmically driven world are readily apparent: who is at fault, for example, when a shopping algorithm recommends harmful products to an individual with suicidal ideation (Twohey and Dance)?

It may be suggested that, though the individuals employing these algorithms cannot necessarily scrutinize the decision-making process and demand answers the way they could from a human agent, nevertheless external actors criticize the “choices” of algorithms constantly, such as when a self-driving car veers off course and causes unnecessary harm. The issue at hand in such a case is one of standards of decision-making. Presumably a given profession adopted a given algorithmic decision-maker on the belief that its decisions would be more reliable, productive, or efficient than our own unaided judgment. If we accept an algorithm as the arbiter of a decision in a particular field or application, on what basis will we question it? If we cannot see the process that led to the machine making a decision, nor question it about the basis of its choice, how can we provide the sort of sympathetic feedback Smith thought so essential? If our own sympathetic judgment is to be mounted against that the calculative rationality of the algorithm we have adopted, say, in medicine or law, of what use was the algorithm?

This problem of inscrutability and absence of criticism is not lost on the companies who proliferate these technologies. Inscrutability is the subject of an entire section of a recent Google white paper on the subject of AI governance (Webb and Chou 2019). Google’s policy analysts ask how someone can meaningfully disagree with the decisions of a complex mathematical process that they can neither understand nor observe (Webb and Chou 2019, 12). Those perhaps best equipped to make informed judgments in critical fields such as medicine and law are, Webb and Chou suggest, increasingly asked to sacrifice their judgment to automated processes developed by third parties. Since this choice is not meaningfully made by any one individual, it cannot be judged and critiqued in the same way Smith imagined social and economic choices being critiqued (see TMS III.3.38 and WN I.x.c.12). People are apparently increasingly being deprived of opportunities to exercise judgment; on Smith’s account, this will lead to a degradation of the humanity and freedom of those so effected. If the applications of these new technologies are as ubiquitous as Helbing et al. and the researchers at Google suggest, the accompanying harms illuminated by Smith’s moral theory will be similarly ubiquitous. As David Schmitz writes of Smith’s account of freedom, “We face an abiding risk of waking some day to find that we have been shackled by crony capitalists, or by ‘men of system.’ We also face risks from within— risks that we will not wake up, and will not realize we have been shackled by social pressure”(Schmitz 2016). Smith’s account of technology then helps us see that this risk, particularly the risk of not “waking up” to our own dependence, is heightened through the proliferation of automated technologies.

Think of an example as mundane as the proliferation of commercially available GPS devices for personal navigation. Reliance on turn-by-turn directions, readily available in almost all cases, substantially changes how one travels and thinks about roads and spaces. Many find themselves unable to competently navigate areas familiar or unfamiliar without the aid of these devices (Dahmani and Bohbot 2020). This relatively trivial example suggests some truth to Smith’s fear: when we are trained to be dependent on others, rather than trained to make our own judgments independently, we degrade in our faculties in some way. The consequences of overdependence on GPS technology may be small, but the cumulative effect of reliance on complex technology may be great indeed. This paper, then, is not a comprehensive argument, suggesting that contemporary device-users are or will necessarily be condemned to mental mutilation and utter dependence. It is rather opening the door to new and useful criticisms of new technologies on the groundwork that Smith has laid. What Smith offers us then are particular criticisms that apply to certain uses and manifestations of new machines; how we can mitigate the harms Smith helps us identify is somewhat less clear. ¹²

Conclusion

New technologies span the gap between labor and leisure, allowing large portions of the lives of contemporary consumers to be dictated by choices other than their own. Everything from navigation to entertainment to shopping to socializing is based on predictive algorithms that increase efficiency while giving the illusion of choice and the feeling of the exercise of judgment. The potential consequences of these developments are both known by those driving this change and are illuminated in their scope by Smith’s moral psychology and analysis of the division of labor and technology. The utility of Smith’s thought on the development and practice of judgment and freedom is most obvious and relevant to conversations surrounding algorithmic artificial intelligence technologies which quite literally substitute the decisions of algorithm for those of an individual. Bringing Smith into our contemporary technological conversation brings with him useful categories and concepts for assessing, criticizing, and addressing new technological developments. Without this key conception of freedom, many technological criticisms fail to fully unpack the depth of the dangers presented to liberal freedoms by our use of new technologies.

We then come to see the utility of Smith’s theory of freedom to contemporary technological commentary by uniting the visions of judgment in both moral and economic arenas across Smith’s TMS and WN. Smith’s fears of overspecialization may not match our observations of the trends of current technological developments, although we would do well to remember many workers globally still find themselves in productive situations Smith would call mentally mutilating. In assessing our new technological developments, however, it is Smith’s description of the importance of meaningful social feedback, self-criticism, and refinement of judgment that exposes new threats beyond what Smith may have been able to foresee. Where new applications of algorithmic, decision-making machines fall afoul of Smith’s conception of freedom is identified cleanly by Helbing et al., researchers at Google, and others: algorithms are pervasive in their reach, and inscrutable in their mechanisms, widely substituting the decisions of a machine for the decisions of the individual. Training consumers to be reliant on the choices of “others” in the form of algorithms could be just as threatening to liberty, then, as training consumers to be reliant on the choices of paternalistic regulators or economic planners. Contemporary machines operate much like paternalistic restrictions on the freedom of the worker that Smith explicitly criticized, while sharing in the harms he articulates in his description of mental mutilation. In both cases, the scope of the exercise of judgment is narrowed, and opportunities for refining that judgment are limited.

I wish to conclude, then, with two suggestions for further research along these lines. First, within political theory, there is significant room for discussion of the political theory of algorithmic decision-making generally. Those working in AI governance and policy clearly present normative concerns, and theorists stand equipped to speak to those concerns and provide insight into these practical conversations. ¹³ Second, in looking to further applications of Smith in technological conversations, it seems that Smith’s theory of spectator-based, social refinement of judgment might fruitfully be applied to new online social spaces; the challenges faced to the development of sympathy and the faculty for judgment in these relatively new social environments deserve careful treatment, and Smith’s detailed mechanisms outlined in TMS could prove valuable indeed.