Most tech revolutions made work worse for employees. AI could be the exception
this+that ● Covered by 2 sources
Opinion — commentary, not a factual news event.
AI is speeding up work again, and history says that usually means more work, not less free time. But this piece argues AI might break the pattern by actually doing tasks, not just accelerating them.
Every big leap in office technology has followed the same script: tools promise freedom, and then someone quietly raises the quota. The PC killed the secretarial pool because people started typing their own memos. Email replaced the mailroom, and smartphones erased the line between the office and everywhere else. None of that spare time went back to the worker. Between the early 1980s and 2015, the expectation for a typical knowledge worker didn't just double the speed of the job — it doubled the workload, from something like 10 units of output a week to 20.
Researchers have a name for why this keeps happening: the autonomy paradox. Give someone a laptop and a mobile inbox so they can work from a beach, and within a few years the beach becomes just another desk. Nobody sits down and decides this. The savings simply get absorbed by rising expectations, and the group converts individual flexibility into a norm of constant availability. It's a pattern that shows up whether you're studying dictation machines or Slack notifications.
AI looks different on paper because, unlike the PC or email, it doesn't just move information faster — it actually produces some of the output. An agent can draft the email, generate the quote, and hit send while its human is out for a run. McKinsey estimates that by 2030, AI could hand knowledge workers back roughly 30 percent of their day, something like two and a half hours. That's not nothing, but early signs are mixed: many workers report AI has made their jobs more intense, not less, as responsibilities blur and people absorb tasks once handed off to specialists.
The better precedent might be desktop publishing, not the PC. When anyone could suddenly make a newsletter with clip art and fancy fonts in the early 1990s, printshops braced for extinction. Instead, the tools raised the bar for what counted as good design, and the ranks of professional designers in the U.S. roughly doubled between 1983 and 2001, climbing from 393,000 to 788,000, even as typesetting jobs vanished. Desktop publishing didn't accelerate the treadmill — it shifted the competition toward taste and craft.
The question for AI is whether reclaimed hours turn into more widgets churned out at the same quality, which just floods markets and drives prices down, or whether they turn into better widgets. Proximity research offers a clue about where the upside might come from: MIT's Thomas Allen found that collaboration between coworkers drops off sharply past about 50 meters, and Bell Labs was literally designed around forcing people from different disciplines to bump into each other in hallways. If AI actually frees up two and a half hours a day, the winners may be the companies that spend that time on walks, hallway conversations, and messy collaboration rather than cramming in a few more tasks.
My take
The autonomy paradox is the real headline here, not AI itself — every efficiency tool since the fax machine has been quietly rerouted into higher quotas instead of shorter days, and there's no reason to assume management culture has suddenly grown a conscience. If AI actually frees up real hours, the companies worth working for will be the ones that spend that time on better work instead of more of it, and everyone else will just find a new way to fill the tank.
Read more about this at: this+that
