The corporate office and its paperwork have a way of turning even ostensibly good jobs – the kind that provide decent salaries and benefits and take place behind ergonomic keyboards in climate-controlled comfort – into soul-sucking drudgery.

In 2013, the now-deceased radical anthropologist David Graeber gave the world a distinct way to think about this problem in an essay called On the Phenomenon of Bullshit Jobs. This anti-capitalist polemic by the man who had helped coin Occupy Wall Street’s iconic “99%” slogan went viral, seemingly speaking to a widely felt 21st century frustration. Graeber developed it into a book that delved deeper on the subject.

He suggested that economist John Maynard Keynes’ dream of a 15-hour workweek had never come to pass because humans have invented millions of jobs so useless that even the people doing them can’t justify their existence. A quarter of the workforce in rich countries sees their jobs as potentially pointless, according to a study by Dutch economists Robert Dur and Max van Lent. If workers find the labour dispiriting, and the work adds nothing to society, what’s the argument for keeping these jobs?

The stakes of that question have heightened as artificial intelligence hurtles forward, bringing with it the spectre of job displacement. A recent estimate by Goldman Sachs found that generative AI could eventually automate activities that amount to the equivalent of some 300 million full-time jobs globally – many of these in office roles like administrators and middle managers.

When imagining a future where technology replaces human effort, we tend to think in two extremes: as a productivity boon for businesses and a disaster for the humans who will become obsolete.

There is a possibility that lies somewhere between these scenarios, however, in which AI kills off some jobs that workers themselves deem meaningless, and even find psychologically degrading. If it did, would these workers be better off?

Flunkies, goons and box tickers

The way researchers talk about AI can sometimes sound like a human resources manager evaluating the bushy-tailed summer intern: shows tremendous promise! It is evident that AI can do quite a lot – mimicking Shakespeare, debugging code; sending emails, reading emails – though it’s not at all clear how far it will go, or what consequences that will bring.

Robots are adept at pattern recognition, which means they excel at applying the same solution to a problem over and over: churning out copy, reviewing legal documents, translating between languages. When humans do something ad nauseam, their eyes might glaze over, they slip up; chatbots don’t experience ennui.

These tasks tend to overlap with some of those discussed in Graeber’s book. He identified categories of useless work including “flunkies,” who are paid to make rich and important people look more rich and important; “goons,” who are hired into positions that exist only because competitor companies created similar roles; and “box tickers,” which are, admittedly, subjective. Some economists, trying to make the designation more useful, have sharpened it: jobs that workers themselves find useless, and which produce work that could evaporate tomorrow with no real effect on the world.

An obvious candidate for “flunky” automation is the executive assistant. IBM already allows users to build their own AI assistants. On Gmail, writers no longer have to compose their own responses, because auto reply generates choices like “yes, that works for me”. AI is even promising to take over personal logistics: AI startup Duckbill uses a combination of AI and human assistants to knock out rote to-do-list items entirely, from returning purchases to buying a child’s birthday present – chores that might have once been shunted to front-desk girls in the Mad Men era.

In other words, when it comes to administrative work, AI has already arrived. That reality crashed down on Kelly Eden, 45, a writer who has for years financially supplemented her magazine writing with administrative work like drafting emails for businesspeople. One of Eden’s most reliable clients owned a chocolate company and paid her 50 cents a word to draft his emails. This year, the chocolatier called to say he would start using ChatGPT instead. Eden was hit with the painful realisation that she needed a backup plan for the work supporting her most fulfilling pursuits.

Telemarketing, another area that AI is overtaking, qualifies as a “goon” job in Graeber’s assessment, because workers often sell products that they know customers don’t really want or need. Chatbots are good at this because they don’t care whether the task is fulfilling, or if customers are surly. Call centres like AT&T’s are already using AI to script calls with customer service representatives, which has made some of those representatives feel as if they are training their own replacements.

Software engineering jobs can veer into “box ticking” territory. That was what Wang felt when he wrote lines of code that didn’t go live. As far as he could tell, the only function this work served was to help his bosses get promoted. He is keenly aware that much of this work could be automated.

But whether or not these jobs provide a sense of existential purpose, they do provide reliable salaries. Many of the meaningless jobs that AI could overtake have traditionally opened up these white-collar fields to people who need opportunities and training, serving as accelerants for class mobility: paralegals, secretaries, assistants. Economists worry that when those jobs disappear, the ones that replace them will bring lower pay, fewer opportunities to ascend professionally and – even less meaning.

“Even if we take Graeber’s view of those jobs, we should be concerned about eliminating them,” said Simon Johnson, an economist at the Massachusetts Institute of Technology. “This is the hollowing out of the middle class.”

A ‘species-level identity crisis’

It’s nearly impossible to imagine what the labour market will look like as AI improves and transforms our workplaces and our economy. But many workers booted from their meaningless jobs by AI could find new roles, ones that emerge through the process of automation. It’s an old story: Technology has offset job losses with job creation throughout history. Horse-drawn carriages were replaced by cars, which created jobs not just on auto assembly lines but also in car sales and gas stations. Personal computing eliminated some 3.5 million jobs, and then created an enormous industry and spurred many others, none of which could have been fathomed a century ago, making clear just why Keynes’ prediction in 1930 of 15-hour workweeks seems so far out of reach.

Kevin Kelly, a Wired co-founder who has written many books on technology, said he was somewhat optimistic about the effect AI would have on meaningless work. He said he believed that partly because workers might begin probing deeper questions about what made a good job.

Kelly has laid out a cycle of the psychology of job automation. Stage 1: “A robot/computer cannot possibly do what I do.” Stage 3: “OK, it can do everything I do, except it needs me when it breaks down, which is often.” Skip ahead to Stage 5: “Whew, that was a job that no human was meant to do, but what about me?” The worker finds a new and more invigorating pursuit, leading full circle to Stage 7: “I am so glad a robot cannot possibly do what I do.”

It’s demoralising to realise that your job can be replaced by technology. It can bring the pointlessness into sharp relief. And it can also nudge people to ask what they want out of work and seek out new, more exhilarating pursuits.

“It might make certain things seem more meaningless than they were before,” Kelly said. “What that drives people to do is keep questioning: ‘Why am I here? What am I doing? What am I all about?’”

“Those are really difficult questions to answer, but also really important questions to ask,” he added. “The species-level identity crisis that AI is promoting is a good thing.”

Some scholars suggest that the crises prompted by automation could steer people toward more socially valuable work. Dutch historian Rutger Bregman started a movement for “moral ambition” centred in the Netherlands. Groups of white-collar workers who feel that they are in meaningless jobs meet regularly to encourage one another to do something more worthwhile. (These are modelled on Sheryl Sandberg’s Lean In circles.) There’s also a fellowship for 24 morally ambitious people, paying them to switch into jobs specifically focused on fighting the tobacco industry or promoting sustainable meats.

“We don’t start with the question of, ‘What is your passion?’” Bregman said of his moral ambition movement. “Gandalf didn’t ask Frodo, ‘What’s your passion?’ He said, ‘This is what needs to get done.’”

What will need to get done in the AI era is likely to veer less toward sustainable meat and more toward oversight, at least in the immediate term. Automated jobs are especially likely to require “AI babysitters,” according to David Autor, an MIT labour economist focused on technology and jobs. Companies will hire humans to edit the work that AI makes, whether legal reviews or marketing copy, and to police AI’s propensity to “hallucinate”. Some people will benefit, especially in jobs where there’s a tidy division of labour – AI handles projects that are easy and repetitive, while humans take on ones that are more complicated and variable. (Think radiology, where AI can interpret scans that fit into preset patterns, while humans need to tackle scans that don’t resemble dozens that the machine has seen before.)

But in many other cases, humans will end up mindlessly skimming for errors in a mountain of content made by AI. Would that help relieve a sense of pointlessness? Overseeing drudge work doesn’t promise to be any better than doing it, or as Autor put it: “If AI does the work, and people babysit AI, they’ll be bored silly.”

Some of the jobs most immediately at risk of being swallowed up by AI are those anchored in human empathy and connection, Autor said. That’s because machines don’t get worn out from feigning empathy. They can absorb endless customer abuse.

The new roles created for humans would be drained of that emotional difficulty – but also drained of the attendant joy. Sociologist Allison Pugh studied the effects of technology on empathic professions like therapy or chaplaincy, and concluded that “connective labour” has been degraded by the slow rollout of technology. Grocery clerks, for example, find that as automated checkout systems come to their stores, they’ve lost out on meaningful conversations with customers – which they understand managers don’t prioritise – and now are left mostly with customers exasperated about self checkout. That’s partially why Pugh fears that new jobs created by AI will be even more meaningless than any we have today.

Even techno-optimists like Kelly argue that there’s a certain inevitability to meaningless jobs. After all, meaninglessness, per Graeber’s definition, is in the eye of the worker.

And even beyond the realm of Graeber’s categories of pointless work, plenty of people have ambivalent relationships with their jobs. Give them enough hours and then years clocking in to do the same things, and they might start to feel frustrated: about being tiny cogs in big systems, about answering to orders that don’t make sense, about monotony. Those aggrieved feelings could crop up even as they jump into new roles, while the robot cycles spin forward, taking over some human responsibilities while creating new tasks for those who babysit the robots.

Some people will look for new roles; others might organise their workplaces, trying to remake the parts of their jobs they find most aggravating, and finding meaning in lifting up their colleagues. Some will search for broader economic solutions to the problems with work. Graeber, for example, saw universal basic income as an answer; OpenAI’s Sam Altman has also been a proponent of experiments with guaranteed income.

In other words, AI magnifies and complicates the social issues entwined with labour but isn’t a reset or cure-all – and while technology will transform work, it can’t displace people’s complicated feelings toward it.

Wang says he certainly believes that will hold true in Silicon Valley. He predicts that automating pointless work will mean engineers get even more creative about seeking out their promotions. “These jobs exist on selling a vision,” he said. “I fear this is one problem you can’t automate.”

This article originally appeared in The New York Times

Written by: Emma Goldberg

Photographs by: Pablo Delcan

©2024 THE NEW YORK TIMES