AI and the Ethics of Miserable Workplaces

AI is changing the workplace, and not always for the better. In September, European Uber drivers launched a class action suit against the company regarding an allegedly invasive AI pay-setting system. Data labeling, where humans perform actions such as flagging objects in pictures or annotating text to facilitate AI learning, has emerged as a prominent field of work and is often repetitive and poorly paid. A German study recently found that as digitalization (including AI) comes for the workplace, health risks actually increase among workers primarily performing physical tasks. Even among white-collar digital workers, it was found they now spend an average of 6.4 hours per week “botsitting,” or making AI work usable. How AI overall affects worker satisfaction and well-being is not yet clear, but it may be that the concerning effect of AI is not so much unemployment as tedium, misery, and loss of agency.

This invites a broader ethical discussion: What should we think about the character of the workplace? Should we even care?

For the most egregious workplace practices, such as unsafe working conditions, physical punishment, or extreme surveillance, a government’s general mandate to protect its populace may suffice. Strong worker’s rights may also help workers secure better working conditions. But what about work that is not so much straightforwardly abusive, as unpleasant? Is there an ethical interest here? Granted, not everyone finds the same things odious, but assuredly some workplace characteristics, such as repetitive activities, micromanagement, lack of agency, and dealing with human excrement are found unpleasant by almost everyone.

On one vision of work, it is merely something done for compensation. As the saying goes, “If it were fun, they wouldn’t call it work.”

Adam Smith, that founding thinker in economics, is likely the most important in this lineage. He saw work as something unpleasant, requiring the sacrifice of leisure and liberty. Generally, the more unpleasant the labor the more it should be rewarded. This does not mean that the most unpleasant work should be the most highly paid work in an absolute sense — considerations such as skill and bargaining power still apply. But if two jobs are otherwise similar, the more unpleasant should be expected to offer greater compensation to attract employees.

While Smith did actually have concerns about the character of work and how it impacted workers, his thinking leads naturally to an economic approach where there is no special concern about what work is like. Expected misery or satisfaction are simply features of a job to be weighed, and employees can take it or leave it at a given level of pay.

Unsurprisingly, there is complexity here. First, such an economic approach assumes that humans can crunch the numbers about how much they value pay versus working conditions. But this is easier said than done. You have to accept a job before you experience what it is like. Second, work can change such that what was once satisfying no longer is. Third, even decidedly miserable work can be hard to leave, whether because of competing obligations, lack of alternatives, or job market dysfunction.

But these are more concerns about market function, bargaining power, and labor market friction than conditions of work as such.

However, even within a market-based system workplace satisfaction is not an unimportant question. The ostensible appeal of such a system is based on its ability to satisfy wants through producing goods and providing the means for workers to pay for those goods (it is a tricky question how well the economic satisfaction of wants maps onto something like happiness or well-being). But this gain must be balanced against the satisfaction stemming from work itself. If work becomes sufficiently miserable, the ethical justification for the system goes under.

On another analysis, work is somehow special. This does not imply that work should necessarily be fun, but rather that there is a general ethical interest in ensuring work is meaningful or dignified.

The German philosopher Georg Wilhelm Friedrich Hegel valued work for allowing us to see ourselves in the world, especially through craft. It is critical to an awareness of our ability to shape the world and the effort that may require. Karl Marx built on Hegel’s ideas, but argued that under capitalism in which we are working for a wage, rather than see ourselves in our work, workers become alienated from the products of their labor, their labor itself, and ultimately the human capacity to meaningfully shape the world. Even philosophers who do not follow Marx in his specific critique of wage labor, such as Andrea Veltman, contend that work can be an important source of purpose, discipline, esteem, and other positive values. And indeed, psychological findings generally support the claim work is important beyond compensation – though as always it is hard to say how much these results depend on current cultural attitudes towards work as opposed to something deep about humans.

Across these analyses work is not merely something we do in exchange for compensation. The character of work matters greatly and it becomes a major political and ethical question how to organize society to secure it. This can involve much more than minor improvement to the workplace. It might even require the revolution Marx foretold.

One problem that emerges concerns work that almost no one wants to do, but that must be done. Whatever one’s philosophy of work, sewers still need to be unclogged. Scholars such as Jan Kandiyali have proposed a sharing system, where such work would be distributed or rotated to prevent it from falling on specific people. AI and robotics are also potential solutions to such work. The tongue-in-cheek description “Fully Automated Luxury Communism” is used to describe a society where technology and automation is used to reduce scut work and facilitate leisure.

Another motivating reason for caring about the quality of the workplace, is that work does not necessarily stay in the workplace. Even Adam Smith worried about how the psychological effects of work could bleed into who we are, especially repetitive work which did not involve creativity, reflection, or judgment and resulted in workers becoming “as stupid and ignorant as it is possible for a human creature to become.” Rather than stripping thought from the workplace by performing highly structured tasks, AI instead “thinks” for people, but the effect may be the same. 60% of workers worry about AI eroding their skills according to a recent IBM Institute of Business Value survey.

To be clear, AI does not need to have a specific effect on labor. It is an implementation question. In 2023, with the launch of Microsoft Co-pilot, CEO Satya Nadella claimed, “This new generation of AI will remove the drudgery of work and unleash creativity.” This may very well be true, at least for some people in some contexts. In any event, there remains a practical question here. The current incentive for owners is to increase efficiency with improving the character of the workplace at most a side benefit. Either regulation or labor power may be needed to push AI integration in a direction that supports workers. But there is also a philosophical question: What should work do? And how should the ideal workspace be constructed?