On labor and AI
by Matt May
Happy Labor/Labour Day/bank holiday to you. Seems like a good day to talk about the elephant in the room again.
The predominant storyline around labor, of course, is AI, the greatest potential disruptor to the workplace since the Industrial Revolution. There’s been no end to the predictions of how this stuff will transform society, but in my opinion, far too little about why we need it to, much less whether it should.
To listen to the CEOs of large language model (LLM) companies, though, you’d think it was already such a great success that they’re generating trillions of dollars in profits, and we just need to figure out how to parcel it out. (Sidebar: maybe start by paying for the content you stole? Just a thought.) OpenAI CEO Sam Altman commissioned a study to determine how to roll out a universal basic income, because as we all know, our political system depends on rich guys deciding unilaterally how (or whether!) their profits are to be redistributed. He then came back with the brilliant idea that OpenAI should give out… a monthly ration of access to their tools. Hope your kid can eat compute credits!
More recently, Nvidia’s CEO, Jensen Huang, said AI implementation could “probably” lead to a transition to a four-day work week. “Every industrial revolution leads to some change in social behavior,” he said, handwavingly, in reference to how we used to work six or seven-day weeks, and then apparently something magical happened, and now we work only five. (Well, except for in China, and now some parts of Silicon Valley, where “996” is shorthand for working from 9am to 9pm, six days a week.)
For those of you whose US social studies memory is hazy, the five-day work week didn’t just happen, and it wasn’t just a gift from all the captains of industry, having gotten together and decided they were rich enough and needed to spread the wealth out of the goodness of their hearts. In fact, it was several decades of labor union organizing and strike activity that led many companies to offer eight-hour workdays—and incidentally, also led Congress in 1894 to dedicate a federal holiday in September to honor workers.
Back to our modern oligarchs, and their claims of a worker-friendly AI paradise. Exactly where is this four-day week going to come from? Worker-friendly governments are few and far between, worldwide. Nvidia isn’t exactly going to turn its GPUs off three days a week to enforce it.
And just think about it for like two seconds: if you tell a CEO that their employees are getting 40 hours of work done in 32, what are they more likely to do?
- Institute a 4-day/32-hour week at the same compensation
- Cut one day and 20% of compensation
- Keep 40 hours and demand a 25% increase in productivity
- Fire 20% of employees
Here’s the thing: if you are not actively organizing for 1. to happen, then one or more of the others will.
When you look at it from the AI CEOs’ perspective, the hard sell is necessary. The LLM companies are deeply underwater. OpenAI is spending over $2 for every $1 they take in, and is expected to continue to lose an eyewatering $44 billion by 2029. The only way OpenAI et al. can survive long enough to be the hegemon they see themselves becoming is to make a generative-AI takeover sound inevitable. They talk about revenue in terms of trillions, not because they have a path to it, but because it’s a number that investors and nation-states cannot ignore. Last week, it came out that they tried to sell the UK a blanket ChatGPT license for £2 billion a year. Honestly, pitching a government on what amounts to a universal tax to support your money-losing endeavor is giving more desperation than inevitability.
Nvidia is supplying the picks and shovels right now, so while they’re highly profitable (at $4.2 trillion market cap, they’re the most valuable company ever), they too depend on grandiose predictions of data centers full of their chips to justify it. If the AI sector crashes, Nvidia may not go under, but it could easily lose 80% of its value. They need you to believe it’s inevitable, too.
But we also need to realize that these guys don’t care about labor. They didn’t become billionaires on the promise of augmenting labor. The true promise, the one that’s worth trillions to other CEOs, is in eliminating it. This is why you can’t take any of their promises about the office of tomorrow seriously. In reality, while they’re droning on and on about “artificial general intelligence”—that is, the imaginary future where AI is smarter than humans—95% of AI projects are failing, to deliver value, with most of the investment going to perceived low-value tasks, like sales and marketing. (Worth noting: if your boss thinks you can be fully replaced by the bullshit machine, it’s because they think your job is bullshit.) Not only can’t they replace us, they’re siphoning off money from functional businesses that could actually be growing, and employing more of those pesky humans, if they weren’t pissing their money away on projects that fail 19 times out of 20.
Anyway, what are we all supposed to be doing when AI runs everything? At least in the US, you basically need a job for health care to be remotely affordable. Jobs also subsidize other businesses: for example, employees don’t just need laptops, but Office365 licenses and single-sign-on apps and payroll software seats, and on and on. A contracting labor force is a recipe for recession, as the pool of money floating from company to company spirals downward. And of course, fewer of us will be getting paid, which means we can’t afford things like food, shelter and Labubus.
There’s a certain isolating effect to being super-rich that makes ideas like “that’s okay, you can just have an AI launch your own profitable company with your ChatGPT Stamps™” sound plausible, instead of like the dumbest thing anyone has ever thought, ever. So it should follow that, you know, maybe large numbers of us should come to a common conclusion that these bubble-boy charlatans shouldn’t just get to decide that the world’s economy (not to mention its ecology) should revolve around them. For example, if I’m right, and most if not all of these companies are going to collapse, they should not get one dime in bailouts. Let them die. Let the investors lose their shirts. Use that money to help everyday folks survive the aftereffects.
Bottom line: if we want the kind of protections that our ancestors secured through blood and sweat, we’re going to need to organize to do that. I worry that too many people in tech are afraid to rock the boat, particularly while their salaries depend on the AI bubble not bursting. I am hoping people start to think about their own futures, and how much time they have left before they’re the ones on the chopping block. These companies, and their products, are not your friends.
Office hours
I keep my calendar open on Thursdays for people who want to talk about working in DEI roles in tech, especially given, you know, all this.
https://calendly.com/practeq/officehours
That’s it. Have a good (hopefully holiday-shortened) week.