The Fault Lies not in AI

Updated: Trump moved against Anthropic to break or bend it to his will. Meanwhile, the panic about AI taking jobs misses the point — capital has used technology to strip workers of power for decades. The real threat was never artificial intelligence.

but in our Capitalism

Update: Anthropic Lawsuit. Just as Trump fought hard to get TikTok sold to a friend, he moved against Anthropic in an unprecedented way to either break the company or bend it to his will. While encouraging to see users leave the compliant ChatGPT for Anthropic over Department of Defense demands, this conflict and resulting lawsuit against the administration is only the first fight for survival as state resources will be brought to bear.

History grants us the perspective to understand today, if we just remember it. Artificial Intelligence, in all its generative and agentic ways, is a remarkable technology that does require careful consideration. However, fear and panic about AI taking jobs is overblown. We already live in an electronic sweatshop.

Ever since a silicon chip could live in the wild, capitalists worked hard to displace worker power with technology. Barbara Garson wrote on automation and displacement after interviewing workers across industries and sectors for The Electronic Sweatshop back in 1988. Garson described the changes workers experienced after this substantial investment in expert computer systems that removed discretion from their jobs and left decisions to computers. McDonald’s, for example, invested in a fry vat computer to eliminate its last truly skilled position — the french fry cook. It took skill to know when to pull the fries— now every fast food beep you hear is a job a fry cook never had.

But the change was deeper and broader than fast food. Companies and governments invested heavily to capture their workforce’s expertise to then run it on computers to make the next generation powerless to resist— a domestic offshoring of skills into technology lit by a dim green screen.

"Right now a combination of twentieth-century technology and nineteenth-century scientific management is turning the Office of the Future into the factory of the past...The primary targets now are professionals and managers." — Barbara Garson

Garson found the same story repeated from the fry cook to the fighter pilot. Welfare caseworkers could no longer weigh individual circumstances — the computer determined eligibility. Airline booking agents lost the discretion to creatively reroute passengers. Stockbrokers watched their own recommendations replaced by computer-generated portfolios. Worker experience had been mined to build expert systems that later robbed workers of the power their expertise afforded them in the workplace.

When Garson wrote, expert systems were bespoke creations, expensive to build and deploy. As computer technology spread, software like Salesforce or Oracle made it more economical to press down on workplace discretion. But, the important motivation for this change, from the early expert systems till today, is not quite based on what you may think, according to Garson:

I had assumed that employers automate in order to cut costs. And, indeed, cost cutting is often the result. But I discovered in the course of this research that neither the designers nor the users of the highly centralized technology I was seeing knew much about its costs and benefits, its bottom-line efficiency. The specific form that automation is taking seems to be based less on a rational desire for profit than on an irrational prejudice against people.

Higher management wanted orders followed. They saw an optimal business path and people just got in the way— an innovating worker, on balance, is more problem than benefit. So today, thanks to this electronic sweatshop mentality, many jobs in our society are already constrained by computers. AI could actually make those less rigid with a greater ability to parse unusual cases. If AI is any threat, it is simply because it lowers the barrier to entry for building expert systems. But this path is already a well-blazed trail.

A Limit on Change

Do not misunderstand me—there will be upheaval in Silicon Valley. When a non-programmer like me can create a phone app using Claude Code in 4 hours (more on this in the future), the entire software industry for the past 50 years will be turned on its head. But nothing about AI will change work much for the rest of the white collar who never code because there will not be much difference to notice.

Think of the change to personal computers on every desk. People used to dictate letters that a secretary had to type and mail. Now that person just sends an email and secretaries become administrative assistants. More future white-collar work will be "managerial" as in tasking AI to draft presentations then reviewing the product. Administrative assistants will become AI coordinators. But these workers are already in front of a computer at their desks all day, so AI work will not be much different.

AI is as boring as we are

AI working in white-collar realms is possible because most white-collar work is boring. Memos have the same formats, slides break down memos in the same way. Buzzwords change from Six Sigma to Agile, or whatever is trendy, but a marketing plan is a marketing plan; a financial report is a financial report, a bankruptcy filing is a bankruptcy filing.

Consider, for a moment, when you were in fifth grade writing a report on the solar system. You would have a paragraph like this:

Jupiter is the largest planet in our solar system and it’s not even close. It’s so massive that over 1,300 Earths could fit inside it. Jupiter orbits about 484 million miles from the sun, which means it takes sunlight about 43 minutes to reach it. - Claude

At some point, there are only so many ways to say something. Like Claude above, Jupiter has its fixed characteristics so you cannot innovate on that— only list— and there are only so many ways to do that without sounding comical. Move forward and as you write that strategy memo, how much of it is really innovation versus boilerplate introductions, problem statements, transitions and summaries?

AI sounds like us because it catches our necessary everydayness in a tireless and useful fashion. It knows there are only so many ways to write that Jupiter report, or draft a memo, or build a slide deck. This is how AI works for us. If we human beings were constantly innovative, we could not understand one another. As William James said, "Habit is the flywheel of society." We need the routine as a backdrop for innovation— a common ground to ensure we are synchronized.

Society will need policies and training to avoid a tyranny of average that unchecked AI could bring. Jobs will tilt toward apprenticeships in management where humans show entry-level humans, project by project, what AI products are meant to look like and why. Education systems will navigate a new world of plagiarism redefined. These are changes AI may bring. But for most of us, after the personal computers, expert systems, enterprise software, and dumb algorithms from Garson’s day to the dawn of AI, work has already changed forever— decades ago.

My Fear

I remember when Facebook was actually useful for community action, those optimistic days when we talked about how social media technology could promote democracy. Facebook was useful until those in power understood it. Then it became a problem for democracy. Now it is a platform scared of its own shadow as it serves up advertising. Twitter and TikTok, with the same hopes, were aggressively taken off the board by the right-wing. If a technology works for the people’s interest, privilege will break it.

AI could be a tremendous engine of accountability combing through the countless documents, like the Epstein files, that hide privilege’s sins. It can serve as a force-multiplier for small groups trying to take on huge projects requiring powerful data processing and analysis. In short, AI could bring sunlight to all those places privilege does not want us to look.

My fear is the good made possible by AI will be corralled by the privileged if they ever see it as a challenge to their power—as Trump right now tries to do to Anthropic. The powerful will act swiftly to protect themselves by consigning us to a tyranny of a monetized average—AI products, AI music, AI videos—and they will crush us under the weight of the bell curve crest to keep us from ever using the true benefit of this technology against them.

Garson’s book is sadly out of print, but you can find it here. Outside of the occasional clearly dated references (remember Wang word processors), the work is still remarkably relevant nearly 40 years later.