CIGI Logo This story was originally published as “Claims That AI Productivity Will Save Us Are Neither New, nor True” by our friends at CIGI. It has been reprinted here with permission.

As artificial intelligence captures the public imagination, while also exhibiting missteps and failures, enthusiasts continue to tout future productivity gains as justification for a lenient approach to its governance. For example, venture fund ARK Invest predicts that “during the next eight years AI software could boost the productivity of the average knowledge worker by nearly 140%, adding approximately $50,000 in value per worker, or $56 trillion globally.” Accenture claims that “AI has the potential to boost labor productivity by up to 40 percent in 2035 . . . enabling people to make more efficient use of their time.” And OpenAI CEO Sam Altman similarly talks about time savings from menial tasks like emailing.

But what if promises around AI productivity do not necessarily translate into benefits to society?

Today, many fears around AI focus on its potential to replace human workers—whether teachers, lawyers, doctors, artists, or writers. In a 1930 essay, the economist John Maynard Keynes made similar predictions, coining the term “technological unemployment” to refer to “unemployment due to our discovery of means of economising the use of labour outrunning the pace at which we can find new uses for labour.” For Keynes, this was proof positive that “mankind is solving its economic problem.” He predicted that his grandchildren would work fifteen-hour weeks, liberated from economic necessity.

But the recent Global Innovation Index suggests otherwise, raising concerns that “considerable investments in technology, innovation, and entrepreneurship [are] failing to deliver the kind of productivity improvements that improve the lot of people across society.” Indeed, the history of “technological revolutions” paints a different story than the one Keynes anticipated about the benefits of technology-related productivity gains.

Take the example of household appliances in the twentieth century. Sociologist Juliet Schor has examined how so-called labour-saving technologies such as the dishwasher, electric stove, and vacuum cleaner failed to reduce women’s household labour. Instead, “rising standards and expectations of domestic life . . . expanded the hours devoted to cleaning, food preparation, and child rearing.” For example, washing machines and dryers allowed laundry to be done more frequently, “adjusting normative standards of cleanliness to meet efficiencies introduced by these appliances,” Schor notes.

Historian Laine Nooney has chronicled how, despite the personal computer revolution’s promises of efficiency and productivity, people have become chained to their computers to the detriment of the human body. Similar claims were made around how laptops and smartphones would untether us—they haven’t. Indeed, these devices have made it possible to work from anywhere, anytime. Rather than this having a liberating effect, we experience “work metastasizing throughout the rest of life,” as Jenny Odell, author of How to Do Nothing: Resisting the Attention Economy, puts it—a phenomenon that was on heightened display for women and working mothers during the pandemic. In fact, these technologies have so drastically eroded boundaries that some jurisdictions are entertaining right-to-disconnect laws.

And now, argues tech writer Paris Marx, “new technologies like AI are framed as offering us various forms of empowerment and liberation: We’ll be able to work more productively, spend less time doing our chores, and anything we want will be a click or tap away. But those promises never paint an accurate picture of how that tech is transforming the world around us or the true cost of those supposed benefits.”

History has shown us that gains in efficiency or productivity as a result of new technologies rarely liberate those already overburdened in society. Instead, new tech often creates new expectations and norms, heightening standards and the amount of work required to attain them. Known as Parkinson’s law, it’s the idea that “work expands so as to fill the time available for its completion.” We have all experienced how meetings scheduled to last an hour will stretch to fill the time allotted.

Increasingly, our standards exceed human capabilities, both physical and cognitive. Nooney notes that “if computers could change how much data a worker could process, then the human body no longer intervened on profitability with its pesky physiological limits.” Similarly, experts now remark on the benefits of using AI—a worker that doesn’t eat, sleep, or require wages. Just as the computer and smartphone have physically distorted the human nervous system and body, taking a considerable toll on our health and well-being, we are told that we have to adapt to the machines—for example, that we need to develop “machine intelligence”—rather than the other way around.

Not only does new tech often result in more work for people but it also introduces additional kinds of work. Ian Bogost anticipates that AI-powered chatbots such as ChatGPT “will impose new regimes of labor and management atop the labor required to carry out the supposedly labor-saving effort.” Just as computers and software advances have “allowed, and even required, workers to take on tasks that might otherwise have been carried out by specialists as their full-time job,” citing procurement and accounting software as examples, Bogst predicts the “inevitable bureaucratization” of AI.

Who can escape the quantitative and qualitative increase in demands that are likely to result as AI advances? As with earlier technologies, the answer is: likely only those with sufficient economic, social, or political capital. For example, only people who have the privilege and power to refuse or “switch off”—who can afford the “cost of opting out”—may avoid social media altogether. And the benefits of flexibility gained through gig-economy services often accrue at the expense of the growing precarity of workers. Similarly, AI advances that increase productivity are likely to result in increasing the already disproportionate burden on everyone but a privileged elite—a new gig economy is burgeoning around AI labelling and other tasks—unless policies approach productivity claims with a critical eye.

Simply put, the AI productivity narrative is a lie. It holds that by automating tasks, AI will make them more efficient and make us, in turn, more productive. This will free us for more meaningful tasks or for leisurely pursuits such as yoga, painting, or volunteerism, promoting human flourishing and well-being. But if history is any guide, this outcome is highly unlikely, save for a privileged elite. More likely, the rich will only get richer.

Because it’s not technology that can liberate us. To preserve and promote meaningful autonomy in the face of these AI advancements, we must look to our social, political, and economic systems and policies. As Derek Thompson observes in The Atlantic, “Technology only frees people from work if the boss—or the government, or the economic system—allows it.” To allege otherwise is technosolutionism, plain and simple.

Reprinted with permission from the Centre for International Governance Innovation.

Elizabeth M. Renieris
Elizabeth M. Renieris is a lawyer, researcher, and author focused on the ethical and human rights implications of new and advanced technologies.