AI’s productivity theater
AI CAN’T do your job, but an AI salesman CAN convince your boss to fire you and replace you with AI.
When I took my kid to New Zealand with me on a book-tour, I was delighted to learn that grocery stores had special aisles where all the kids’-eye-level candy had been removed, to minimize nagging. What a great idea!
Related: countries around the world limit advertising to children, for two reasons:
1) Kids may not be stupid, but they are inexperienced, and that makes them gullible; and
2) Kids don’t have money of their own, so their path to getting the stuff they see in ads is nagging their parents, which creates a natural constituency to support limits on kids’ advertising (nagged parents).
There’s something especially annoying about ads targeted at getting credulous people to coerce or torment other people on behalf of the advertiser. For example, AI companies spent millions targeting your boss in an effort to convince them that you can be replaced with a chatbot that absolutely, positively cannot do your job.
Your boss has no idea what your job entails, and is (not so) secretly convinced that you’re a featherbedding parasite who only shows up for work because you fear the breadline, and not because your job is a) challenging, or b) rewarding:
https://pluralistic.net/2024/04/19/make-them-afraid/#fear-is-their-mind-killer
That makes them prime marks for chatbot-peddling AI pitchmen. Your boss would love to fire you and replace you with a chatbot. Chatbots don’t unionize, they don’t backtalk about stupid orders, and they don’t experience any inconvenient moral injury when ordered to enshittify the product:
https://pluralistic.net/2023/11/25/moral-injury/#enshittification
Bosses are Bizarro-world Marxists. Like Marxists, your boss’s worldview is organized around the principle that every dollar you take home in wages is a dollar that isn’t available for executive bonuses, stock buybacks or dividends. That’s why you boss is insatiably horny for firing you and replacing you with software. Software is cheaper, and it doesn’t advocate for higher wages.
That makes your boss such an easy mark for AI pitchmen, which explains the vast gap between the valuation of AI companies and the utility of AI to the customers that buy those companies’ products. As an investor, buying shares in AI might represent a bet the usefulness of AI — but for many of those investors, backing an AI company is actually a bet on your boss’s credulity and contempt for you and your job.
But bosses’ resemblance to toddlers doesn’t end with their credulity. A toddler’s path to getting that eye-height candy-bar goes through their exhausted parents. Your boss’s path to realizing the productivity gains promised by an AI salesman runs through you.
A new research report from the Upwork Research Institute offers a look into the bizarre situation unfolding in workplaces where bosses have been conned into buying AI and now face the challenge of getting it to work as advertised:
https://www.upwork.com/research/ai-enhanced-work-models
The headline findings tell the whole story:
- 96% of bosses expect that AI will make their workers more productive;
- 85% of companies are either requiring or strongly encouraging workers to use AI;
- 49% of workers have no idea how AI is supposed to increase their productivity;
- 77% of workers say using AI decreases their productivity.
Working at an AI-equipped workplaces is like being the parent of a furious toddler who has bought a million Sea Monkey farms off the back page of a comic book, and is now destroying your life with demands that you figure out how to get the brine shrimp he ordered from a notorious Holocaust denier to wear little crowns like they do in the ad:
https://www.splcenter.org/fighting-hate/intelligence-report/2004/hitler-and-sea-monkeys
Bosses spend a lot of time thinking about your productivity. The “productivity paradox” shows a rapid, persistent decline in American worker productivity, starting in the 1970s and continuing to this day:
https://en.wikipedia.org/wiki/Productivity_paradox
The “paradox” refers to the growth of IT, which is sold as a productivity-increasing miracle. There are many theories to explain this paradox. One especially good theory came from the late David Graeber (rest in power), in his 2012 essay, “Of Flying Cars and the Declining Rate of Profit”:
https://thebaffler.com/salvos/of-flying-cars-and-the-declining-rate-of-profit
Graeber proposes that the growth of IT was part of a wider shift in research approaches. Research was once dominated by weirdos (e.g. Jack Parsons, Oppenheimer, etc) who operated with relatively little red tape. The rise of IT coincides with the rise of “managerialism,” the McKinseyoid drive to monitor, quantify and — above all — discipline the workforce. IT made it easier to generate these records, which also made it normal to expect these records.
Before long, every employee — including the “creatives” whose ideas were credited with the productivity gains of the American century until the 70s — was spending a huge amount of time (sometimes the majority of their working days) filling in forms, documenting their work, and generally producing a legible account of their day’s work. All this data gave rise to a ballooning class of managers, who colonized every kind of institution — not just corporations, but also universities and government agencies, which were structured to resemble corporations (down to referring to voters or students as “customers”).
Even if you think all that record-keeping might be useful, there’s no denying that the more time you spend documenting your work, the less time you have to do your work. The solution to this was inevitably more IT, sold as a way to make the record-keeping easier. But adding IT to a bureaucracy is like adding lanes to a highway: the easier it is to demand fine-grained record-keeping, the more record-keeping will be demanded of you.
But that’s not all that IT did for the workplace. There are a couple areas in which IT absolutely increased the profitability of the companies that invested in it.
First, IT allowed corporations to outsource production to low-waged countries in the global south, usually places with worse labor protection, weaker environmental laws, and easily bribed regulators. It’s really hard to produce things in factories thousands of miles away, or to oversee remote workers in another country. But IT makes it possible to annihilate distance, time zone gaps, and language barriers. Corporations that figured out how to use IT to fire workers at home and exploit workers and despoil the environment in distant lands thrived. Executives who oversaw these projects rose through the ranks. For example, Tim Cook became the CEO of Apple thanks to his successes in moving production out of the USA and into China.
Outsourcing provided a sugar high that compensated for declining productivity…for a while. But eventually, all the gains to be had from outsourcing were realized, and companies needed a new source of cheap gains. That’s where “bossware” came in: the automation of workforce monitoring and discipline. Bossware made it possible to monitor workers at the finest-grained levels, measuring everything from keystrokes to eyeball movements.
What’s more, the declining power of the American worker — a nice bonus of the project to fire huge numbers of workers and ship their jobs overseas, which made the remainder terrified of losing their jobs and thus willing to eat a rasher of shit and ask for seconds — meant that bossware could be used to tie wages to metrics. It’s not just gig workers who don’t score consistent five star ratings from app users whose pay gets docked — it’s also creative workers whose Youtube and Tiktok wages are cut for violating rules that they aren’t allowed to know, because that might help them break the rules without being detected and punished:
https://pluralistic.net/2024/01/13/solidarity-forever/#tech-unions
Bossware dominates workplaces from public schools to hospitals, restaurants to call centers, and extends to your home and car, if you’re working from home (AKA “living at work”) or driving for Uber or Amazon:
https://pluralistic.net/2020/10/02/chickenized-by-arise/#arise
In providing a pretense for stealing wages, IT can increase profits, even as it reduces productivity:
https://pluralistic.net/2024/01/11/robots-stole-my-jerb/#computer-says-no
One way to think about how this works is through the automation-theory metaphor of a “centaur” and a “reverse centaur.” In automation circles, a “centaur” is someone who is assisted by an automation tool — for example, when your boss uses AI to monitor your eyeballs in order to find excuses to steal your wages, they are a centaur, a human head atop a machine body that does all the hard work, far in excess of any human’s capacity.
A “reverse centaur” is a worker who acts as an assistant to an automation system. The worker who is ridden by an AI that monitors their eyeballs, bathroom breaks, and keystrokes is a reverse centaur, being used (and eventually, used up) by a machine to perform the tasks that the machine can’t perform unassisted:
https://pluralistic.net/2023/04/12/algorithmic-wage-discrimination/#fishers-of-men
But there’s only so much work you can squeeze out of a human in this fashion before they are ruined for the job. Amazon’s internal research reveals that the company has calculated that it ruins workers so quickly that it is in danger of using up every able-bodied worker in America:
https://www.vox.com/recode/23170900/leaked-amazon-memo-warehouses-hiring-shortage
Which explains the other major findings from the Upwork study:
- 81% of bosses have increased the demands they make on their workers over the past year; and
- 71% of workers are “burned out.”
Bosses’ answer to “AI making workers feel burned out” is the same as “IT-driven form-filling makes workers unproductive” — do more of the same, but go harder. Cisco has a new product that tries to detect when workers are about to snap after absorbing abuse from furious customers and then gives them a “Zen” moment in which they are showed a “soothing” photo of their family:
https://finance.yahoo.com/news/ai-bringing-zen-first-horizons-192010166.html
This is just the latest in a series of increasingly sweaty and cruel “workplace wellness” technologies that spy on workers and try to help them “manage their stress,” all of which have the (totally predictable) effect of increasing workplace stress:
https://pluralistic.net/2024/03/15/wellness-taylorism/#sick-of-spying
The only person who wouldn’t predict that being closely monitored by an AI that snitches on you to your boss would increase your stress levels is your boss. Unfortunately for you, AI pitchmen know this, too, and they’re more than happy to sell your boss the reverse-centaur automation tool that makes you want to die, and then sell your boss another automation tool that is supposed to restore your will to live.
The “productivity paradox” is being resolved before our eyes. American per-worker productivity fell because it was more profitable to ship American jobs to regulatory free-fire zones and exploit the resulting precarity to abuse the workers left onshore. Workers who resented this arrangement were condemned for having a shitty “work ethic” — even as the number of hours worked by the average US worker rose by 13% between 1976 and 2016:
https://pluralistic.net/2024/01/11/robots-stole-my-jerb/#computer-says-no
AI is just a successor gimmick at the terminal end of 40 years of increasing profits by taking them out of workers’ hides rather than improving efficiency. That arrangement didn’t come out of nowhere: it was a direct result of a Reagan-era theory of corporate power called “consumer welfare.” Under the “consumer welfare” approach to antitrust, monopolies were encouraged, provided that they used their market power to lower wages and screw suppliers, while lowering costs to consumers.
“Consumer welfare” supposed that we could somehow separate our identities as “workers” from our identities as “shoppers” — that our stagnating wages and worsening conditions ceased mattering to us when we clocked out at 5PM (or, you know, 9PM) and bought a $0.99 Meal Deal at McDonald’s whose low, low price was only possible because it was cooked by someone sleeping in their car and collecting food-stamps.
But we’re reaching the end of the road for consumer welfare. Sure, your toddler-boss can be tricked into buying AI and firing half of your co-workers and demanding that the remainder use AI to do their jobs. But if AI can’t do their jobs (it can’t), no amount of demanding that you figure out how to make the Sea Monkeys act like they did in the comic-book ad is doing to make that work.
As screwing workers and suppliers produces fewer and fewer gains, companies are increasingly turning on their customers. It’s not just that you’re getting worse service from chatbots or the humans who are reverse-centaured into their workflow. You’re also paying more for that, as algorithmic surveillance pricing uses automation to gouge you on prices in realtime:
https://pluralistic.net/2024/07/24/gouging-the-all-seeing-eye/#i-spy
This is — in the memorable phrase of David Dayen and Lindsay Owens, the “age of recoupment,” in which companies end their practice of splitting the gains from suppressing labor with their customers:
https://prospect.org/economy/2024-06-03-age-of-recoupment/
It’s a bet that the tolerance for monopolies made these companies too big to fail, and that means they’re too big to jail, so they can cheat their customers as well as their workers.
AI may be a bet that your boss can be suckered into buying a chatbot that can’t do your job, but investors are souring on that bet. Goldman Sachs, who once trumpeted AI as a multi-trillion dollar sector with unlimited growth, is now publishing reports describing how companies who buy AI can’t figure out what to do with it:
Fine, investment banks are supposed to be a little conservative. But VCs? They’re the ones with all the appetite for risk, right? Well, maybe so, but Sequoia Capital, a top-tier Silicon Valley VC, is also publicly questioning whether anyone will make AI investments pay off:
https://www.sequoiacap.com/article/ais-600b-question/
I can’t tell you how great it was to take my kid down a grocery checkout aisle from which all the eye-level candy had been removed. Alas, I can’t figure out how we keep the nation’s executive toddlers from being dazzled by shiny AI pitches that leave us stuck with the consequences of their impulse purchases.
If you’d like an essay-formatted version of this post to read or share, here’s a link to it on pluralistic.net, my surveillance-free, ad-free, tracker-free blog:
https://pluralistic.net/2024/07/25/accountability-sinks/#work-harder-not-smarter
Image:
Cryteria (modified)
https://commons.wikimedia.org/wiki/File:HAL9000.svg
CC BY 3.0
https://creativecommons.org/licenses/by/3.0/deed.en
