Bernie Sanders’ Worker Dystopia: Never Lose Your Job But Never Get a Raise
If Senator Sanders wants to raise wages, he should focus on the real cause of slow growth, lagging productivity from low capital investment, instead of stymieing AI.
What happens when you combine anti-capitalism with anti-technology? Senator Bernie Sanders’ (I–Vt.) new report on AI and the labor market.
There is so much that is wrong with this report, one hardly knows where to start. One place is the prominent chart that supposedly shows that wages have been stagnant, productivity has grown a fair amount, and profits have gone through the roof.
Although it looks compelling, it’s inaccurate.
The report describes the chart as, “But the resulting economic gains have gone almost exclusively to those at the top. While productivity has risen by 150% and corporate profits have increased more than 370%, real wages have gone down for the average American worker by nearly $30 a week.”
First, it measures total profits but average wages. If it wants an apples-to-apples comparison—which it does not, because this is a piece of political propaganda—it should look at total wage growth. According to the Bureau of Labor Statistics (BLS), total wages increased by 649 percent from 1982 to 2025 (1982 was the earliest year available). Oops! That is more than the total profits increased. And according to the Bureau of Economic Analysis, domestic corporate profit rates were about the same in the late 2010s as they were in the 1970s.
Second, responding to the claim that wages have stagnated while productivity has grown is almost not even worth doing, because it has been rebutted so many times, including by the Congressional Budget Office and in my book with David Moschella, Technology Fears and Scapegoats: 40 Myths About Privacy, Jobs, AI, and Today’s Innovation Economy. But for the sake of clarity, here we go again: When using the right methods and controlling for statistical mismeasurement, there is no divergence between productivity growth and wage growth.
If Senator Sanders wants to help raise the wages of American workers, he should focus on the real cause of slow wage growth: slow productivity growth from low capital investment. Instead, he wants to stymie the most promising technology today to boost productivity and wage growth—AI.
The senator rightly complains about the low federal minimum wage, and I agree. At $7.25 an hour, that is a travesty. Congress should increase it to at least $12 an hour and index it to inflation. As I have written in Democracy: A Journal of Ideas, that would not only boost incomes for lower-wage workers, it would boost productivity overall, helping all American workers.
Senator Sanders’ report then goes on to assert that tech has killed jobs: “Although in the past, jobs lost due to automation led to the creation of new jobs, this trend reversed from 1987 to 2016 when the impact of labor-displacing automation outstripped new job creation.” As ITIF has shown in an analysis of BLS occupational data, in no decade since 1850 has technology directly created more jobs than it has eliminated. However, joblessness throughout most of the period from 1850 to the present has not increased because the savings from productivity-driven gains are reinvested in existing economic functions, allowing jobs such as insurance agents, schoolteachers, nurses, and many others to expand.
Ah, but like the now ubiquitous AI technophobes, the report asserts that this time is different: AI is coming! It quotes several CEOs who speculate that AI will eliminate a bunch of white-collar jobs. But as I wrote in The Wall Street Journal, this is fearmongering speculation that, alas, is highly unlikely to happen. I say alas because the economy desperately needs higher productivity.
The report cites a Goldman Sachs study that estimates AI could lead to a worldwide loss of 300 million jobs. Three hundred million jobs! But wait—that’s less than 10 percent of jobs over ten years. In other words, this tsunami of job destruction would mean the elimination of just 1 percent of jobs a year. According to the BLS, around 6 percent of U.S. workers lose their jobs every year due to downsizing or closures. And the real punchline from the Goldman study is that global productivity is projected to increase, but only by an anemic 1.5 percent over ten years.
Just when you think the report can’t get any worse, it does. Sanders’ staff decided to ask ChatGPT to predict how many jobs it would kill. You can’t make this up. What the heck does ChatGPT know about this? The staff used a fundamentally flawed method, relying on the Department of Labor’s ONET ranking of task complexity for occupations to simplistically assume that the less complex the task, the more likely it is AI will automate it.
This is what Carl Benedikt Frey and Michael Osborne did over a decade ago when they wrongly predicted the loss of 47 percent of jobs. Using this measure, the occupation that Frey and Osborne predicted would have the highest risk of going the way of the buggy-whip manufacturer—insurance underwriters—actually saw employment grow by 16.4 percent from 2013 to the end of 2021. In contrast, the occupation they predicted was the least likely to be automated—recreational therapists—saw a decline of 8.9 percent in jobs.
The reason is simple: machine learning and even large language models are just software. While better than anything we have had before, there are limits to what they can do. But that doesn’t stop the Luddites from crying wolf. As MIT professor and former CEO of Rethink Robotics Rodney Brooks writes, “Misled by suitcase words, people are making category errors in fungibility of capabilities—category errors comparable to seeing the rise of more efficient internal combustion engines and jumping to the conclusion that warp drives are just around the corner.” Beam me up, Scotty.
However, the fundamental flaw in this report is that it channels Karl Marx’s nonsensical notion from almost two centuries ago. The report states, “While this basic analysis reflects all the inherent limitations of ChatGPT, it represents one potential future in which corporations decide to aggressively push forward with artificial labor.” One of Marx’s most egregiously wrong predictions (and there were so many) was that greedy capitalists would use technology to boost productivity, lay off workers, and keep all the gains. The result would be an immiserated proletariat, which is exactly what Senator Sanders is predicting.
Ah, Marx and Sanders forget one thing: Technology does not repeal the laws of competition, which force companies to pass on all the benefits of labor-saving technologies in the form of lower prices and/or higher wages. If they didn’t do that, profit rates would be through the roof, not their current 12 percent or so of national income.
In other words, for Marx and Sanders, companies automate in order to boost profits, and in the process, they immiserate workers. But is the U.S. Postal Service a greedy capitalist enterprise, given that it has used “artificial labor” to boost productivity? (A strange term they offer. Isn’t a tractor or a machine tool artificial labor?) If Sanders could get his way and most industries were nationalized, they all would want to use AI to boost productivity.
But Sanders would oppose that “greed.” He criticizes President Trump for partnering “with Google, OpenAI, and Anthropic to set the table for a large-scale replacement of federal workers with AI.” So, let me get this right: If AI could boost federal government efficiency without reducing quality, which hasn’t been determined yet, Sanders would still oppose that? Because the “shareholders” of the federal government would see massive increases in income while the workers lose their jobs? But the shareholders are taxpayers, who would be able to pay lower taxes, carry a smaller national debt, or see increases in federal spending in other areas, like urban development.
Put another way, Sanders wants a world where no worker ever loses their job—even if it means that their children have the same standard of living as they do. Sounds dystopian.
The report then reviews corporate statements about how some, hopefully many, corporations are using AI to improve efficiency and reduce headcount. But for the authors of the report, this implies fewer jobs. And they, like so many non-economists, commit the “lump-of-labor fallacy”: the idea that there is a limited amount of work to be done and, if a job is eliminated, it’s gone for good. But this fails to account for second-order effects in which the savings from increased productivity are recycled into the economy in the form of higher wages, higher profits, and reduced prices, all of which generate new demand that, in turn, creates other jobs. Some jobs are in new occupations, such as “content creator assistants,” but most are in existing ones that workers will now spend their savings on, like personal trainers, for example. This is why the majority of scholarly studies find no net negative effect of productivity on employment.
And we get to the end of the report to find that this “analysis” is merely a setup to justify Senator Sanders’ progressive social, labor, and economic policy agenda. If he wants to advance that agenda, feel free; some of it ITIF supports, such as improvements to the unemployment insurance system and paid parental leave. But don’t distort reality or the benefits of AI to do that.
Finally, the idea of a robot tax and “ending tax breaks for automation” may be among the worst policy proposals of our era. They would result in even lower wage growth, and we might just end up handing over most of manufacturing to China. Here, like most of the report, it cites fundamentally flawed work. In this case, the Brookings Institution, which in recent years has become decidedly anti-tech: “According to a study from the Brookings Institution, the tax code structurally favors investments in AI and automation over labor.” No, it doesn’t. Companies can expense labor costs in the first year but at least until the passage of the Big Beautiful Bill, generally could not do the same for capital equipment.
I am glad the senator released this report because it highlights perhaps the most important economic issue of the day that needs a robust and open debate: Should workers have protection from automation?
Let’s get this out in the open and debate it fairly. If one believes they should, then the proposal from Senator Sanders and all those similar should be embraced. However, if one believes that this is fundamentally giving in to selfishness and fear, then most of the related proposals should be rejected outright.
The reality is that we now live in a Rawlsian world where it is unethical to have anyone be hurt, whether emotionally or economically. Redistributionist philosopher John Rawls wrote that the goal should be “maintaining conditions and achieving objectives that are similarly to everyone’s advantage.” Following Rawlsian ethics, there can be no automation in an organization because, while 99.999 percent of society will benefit, 0.001 percent will be hurt. Under this ethical framework, progress is virtually impossible, and creative destruction caused by technology would be illegal.
In their growing anti-capitalist animus, both the left and the new right identify the public interest with the interests of workers. For example, the Center for American Progress published an article titled “Will AI Benefit or Harm Workers?” But why workers? Why not children, students, retirees, business owners, or even workers in non-automatable roles? In reality, society is a collection of interests, and if we define the national interest by any particular interest—in this case a worker who might lose their job—we are doomed to economic stagnation. So, no, workers do not deserve protection from automation. What they do deserve is access to robust retraining and support programs to help them transition to new jobs if laid off.