AI, Unions, and the Vast Abyss
People haven't quite grasped that there are no other saviors.
I want to talk about AI and how we may avoid a very bad scenario in the near future. But first, here is a famous chart:
This chart illustrates the basic economic fact that, beginning in the Reagan era, the productivity of American workers has risen much faster than their wages. Workers have not stopped becoming more productive—they have just stopped getting paid for that productivity. The hundreds of billions of dollars that would have gone into workers’ pockets had wages continued to reflect productivity gains have instead been funneled to investors and executives. The reason that workers in the past 40 years became unable to hold onto the top line in that chart is that they lost the power to do so, primarily because of the decline of unions, union density, and labor power as a whole.
It is important to keep in mind today that increased productivity does not necessarily lead to broadly good social and economic outcomes. You have to actively ensure that gains are distributed fairly. NAFTA and international “free trade” are good examples of this. Free trade, in a vacuum? Good! Efficient! Many aggregate productivity gains! Free trade, in practice? The cause of millions of good middle class blue collar jobs disappearing from middle America, paving the way for decades of desolation and opioid addiction. The economic trick that was pulled was saying, “this thing will, in aggregate, create many economic gains,” without mentioning that all of the gains would accrue to the investment class and all of the devastation would accrue to the working class. That “free trade” would cause multinational companies to move high paying American jobs offshore was the most predictable thing in the world. There should have been a serious plan in place to take care of all of those Americans and their families and their cities—a plan that could have been paid for with all of the economic gains that “free trade” was creating. Instead, they just fired away and didn’t much care about the people on the losing side of the equation. Let this be a lesson to us all. You have to build the safety net before you jump off the building.
How Things Work is a reader-supported publication. Consider becoming a paid subscriber if you like it.
And here we are, all of a sudden, in the new era of AI. Writing, photography, audio, video, illustration, and every related field are in imminent danger of “disruption,” in the same sense that NAFTA “disrupted” all of those Midwestern manufacturing workers, straight to poverty. I am not a tech writer; I am concerned with the labor consequences of this issue, and the downstream social and political consequences. Let me suggest a few propositions here that I don’t think should be controversial:
To whatever extent AI technology sucks now—generating stilted writing, unrealistic photos, etc.—it is improving very fast, and yelling about how it’s not as good as human-produced stuff is not a productive way to stave off its uptake by corporations.
Most employers will leap at the chance to replace higher-cost human labor with lower-cost AI labor, to the maximum practical extent possible, for the same reasons that have always driven workplace automation.
It is foolish to believe that the US government is going to pass adequate regulations of AI or produce an adequate social safety net in time to prevent awful social consequences of vast swaths of workers who suddenly find themselves replaced by AI.
That’s were we stand now. It’s coming, corporations are salivating at the prospects of lowering their labor costs, and it is naive to imagine that the US government will save all of the million workers endangered by this. So: What other institutions might we turn to here? The companies that make AI and the companies that can use AI to cut jobs are both going to move as fast as possible. What force exists that can possibly make this a process that moves in a considered, humane, well-managed way, rather than a crude gold rush that leaves millions of us cast aside? I can think of one, and only one: Labor unions. That’s it. Unions are the front line of this issue. Indeed, the only fucking hope.
Who else? NGOs and academia can make a lot of noise about the dangers of AI, but they have little real power to impose any conditions on the companies that will be using the products. The one way that you can impose a condition on a company? A union contract. They are real. They are enforceable. They are the only place that we can reasonably hope to get guidelines for how AI can be used that will actually stick. How else? How else will this happen? There is nothing else.
I am writing this simply to try to make this fact sink in—with the general public, which does not typically associate unions with cutting-edge technological regulations, and, just as importantly, with unions themselves. The Writers Guild of America strike that is happening right now is meaningful for everyone in part because it is the first major union contract in which the use of AI is a serious point of contention. It has the potential to set a very powerful precedent. No matter who you are or what you think about Hollywood screenwriters, the strike is worth supporting for that reason alone.
Ted Chiang has a very correct and intelligible piece in The New Yorker about the implications of AI for capitalism, that creeps right up to this realization, but doesn’t quite grasp it:
As it is currently deployed, A.I. often amounts to an effort to analyze a task that human beings perform and figure out a way to replace the human being. Coincidentally, this is exactly the type of problem that management wants solved. As a result, A.I. assists capital at the expense of labor. There isn’t really anything like a labor-consulting firm that furthers the interests of workers. Is it possible for A.I. to take on that role? Can A.I. do anything to assist workers instead of management?
Unions are the institutions that Chiang cannot put his finger on. His piece is good—I imagine that he doesn’t follow the thread to unions because unions have grown so rare in the private sector that they have fallen out of the minds of much of the general public. Union density in the private sector is six percent. It’s unsurprising that many people don’t think of unions as society-wide power brokers, particularly in the tech industry, which is even less unionized than most.
That’s going to have to change. It’s long been clear that the failure to unionize big tech companies is causing critical damage to the labor movement’s power. (No major tech company is union, although there has been organizing at smaller companies, and among service workers in tech.) And beyond the obvious reasons, a union at a place like Google would also create, in effect, a democratic check on the company’s power—workers could directly influence company policy in a way that the public can’t. Tech unions could also serve as an internal check on the rollout of AI. But realistically, the AI rollout is happening now, and the tech company unions are years away, in an optimistic scenario. It’s up the to the other unions to hold the line.
Screenwriters. Journalists. Actors. Marketing and PR professionals. Architects. Drivers. Logistics workers. Artists. Office assistants. Insurance brokers. Secretaries. Analysts. Finance workers. And a mind-blowing array of white collar employees of virtually every industry. That is a short and incomplete list of people whose livelihoods are threatened today by an unregulated adoption of AI. There are a number of unions in those industries, and, like it or not, they are now responsible for getting a handle on this, for the good of the whole damn country. We’re not talking about some caricature of horse-and-buggy drivers whining about automobiles; we’re talking about the difference between a just transition in which people are taken care of, and a vast and sudden displacement that causes enormous human suffering.
Unions are it. Take a minute to process that. Support them. Unionize your own workplace. And do it soon. We’re already in the middle of this, and we’re just starting to catch up.
I wrote a piece in The Guardian this weekend about why the writers strike is important to everyone. (AI is one reason, but not the only one.) For perspective from actual screenwriters who are on strike, see Josh Gondelman or Max Read. To join a picket line or otherwise support the strike, see here.
“How exactly should AI be regulated” is a separate and equally important question. I’ve written about that as it applies to journalism here, but there is much more to be said.
THANK YOU once again to everyone who has subscribed to How Things Work. If you haven’t yet, why not now? As a fragile human, who loves you much more than AI ever will, I must earn money to buy food in order to survive.