- This topic has 2 replies, 2 voices, and was last updated 2 hours, 19 minutes ago by
.
-
Posts
-
-
Layoffs rock San Francisco’s largest employer as tech giant replaces workers with AI
There are 162.2 million people employed in the United States. Of that, 101.1 million are white collar workers. Personally, I think we will lose between 10 million and 20 million jobs to artificial intelligence.
Women make up 47.1% of the American workforce, but 55% of the 101.1 million white collar workers. How many women will lose their jobs?
AI thinks differently. Here’s OpenAI’s take:
If we define “job losses” as positions eliminated or not replaced because AI can perform enough of the work—rather than people permanently unemployed—I would use about 7 million U.S. jobs over the next five years as a reasonable central estimate, with a plausible range of 4 million to 11 million.That is considerably less apocalyptic than some headlines, but still a very large labor-market event.My five-year estimate: 2026–2031
Measure Conservative Central estimate Aggressive U.S. jobs displaced by AI 4.1M ~7.0M 10.5M % of 162.2M workforce 2.5% 4.3% 6.5% White-collar share ~80% ~85% ~90% White-collar jobs lost ~3.2M ~5.9M ~9.5M Women’s share of losses ~55% ~58% ~62% Women losing/displaced from jobs ~2.2M ~4.0M ~6.5M So if you want three numbers to work with, my best estimate is:
~7 million U.S. jobs displaced by AI by 2031
~85% will be white-collar jobs
~58% will be held by women
These are my synthesis of the research rather than numbers published together by any one institution.
Why I land around 7 million
Goldman Sachs’ August 2025 modeling is particularly useful. It estimates that if today’s AI use cases were spread throughout the economy, 2.5% of U.S. employment would already be at risk of displacement. Its longer-run baseline is 6–7% of employment displaced, with a much wider possible range of 3–14%. Goldman nevertheless expects the unemployment effect to be substantially smaller because displaced workers find other jobs and technology creates new occupations.
On your 162.2-million-worker base, those Goldman benchmarks translate into about 4.1 million jobs at today’s capabilities and roughly 9.7–11.4 million at its 6–7% displacement assumption. Because we are talking about only five years, not complete long-run AI adoption, I would put my midpoint below that latter figure: approximately 7 million positions.
McKinsey independently estimates that nearly 12 million Americans could need to change occupations by 2030, although importantly, that number reflects automation and other structural changes and is not equivalent to 12 million AI layoffs. McKinsey sees particularly large declines in office support and customer service.
And we should keep the forecast humble. As of June–July 2026, both Yale’s Budget Lab and Stanford researchers say there still isn’t evidence of a large aggregate AI employment shock. Yale finds no clear relationship yet between AI usage and overall employment or unemployment, while Stanford sees the clearest negative effects among younger workers in AI-exposed occupations rather than across the entire labor market.
Why I think roughly 85% will be white collar
This is where AI differs dramatically from the great automation waves of the 20th century.
Goldman’s highest-risk occupations include computer programmers, accountants and auditors, legal assistants, administrative assistants, customer-service representatives, telemarketers, proofreaders/copy editors, and credit analysts. That’s an overwhelmingly white-collar list.
Brookings reaches much the same conclusion. The greatest GenAI exposure is concentrated in computer, management, engineering, business/financial and administrative work, while current generative AI has relatively little direct effect on male-dominated physical occupations absent major advances in robotics.
So although white-collar workers constitute only about 62% of the workforce under the broad definition we used earlier, I estimate they will constitute approximately 80–90% of AI displacement during the next five years.
My midpoint: 85%.
That would mean roughly 5.9 million of the 7 million displaced positions are white collar.
And women? This is especially interesting.
Women are only about 47% of all U.S. workers, but they are disproportionately concentrated in some of the occupations most susceptible to AI.
Brookings calculates that 36% of female workers are in occupations where GenAI could save at least half the time required for tasks, versus only 25% of male workers. It specifically identifies administrative and clerical occupations as both highly automatable and heavily female.
If you combine those exposure rates with the gender composition of the workforce, women constitute roughly 56% of the highly AI-exposed population. And actual displacement could tilt somewhat further toward women because occupations such as secretaries, administrative assistants, clerks, bookkeepers and customer-service workers have both high automation potential and high female representation. McKinsey, for example, projects substantial contraction in office-support and customer-service employment, where women are heavily represented.
A particularly striking 2026 Brookings analysis identified 6.1 million workers who combine very high AI exposure with low ability to adapt to displacement; 86% are women. That does not mean 86% of all AI job losses will be women—it identifies the especially vulnerable subset—but it tells us where some of the hardest displacement is likely to fall.
Therefore I’d put the female share of all AI-displaced U.S. workers at roughly 55–62%, with 58% as my working estimate.
One major qualification
I would not write:
“AI will cause 7 million Americans to become unemployed.”
That’s probably wrong. I would write:
“AI could eliminate or displace roughly 7 million existing U.S. positions over the next five years, even as the economy creates new jobs and many displaced workers move into other occupations.”
The mechanism may also be less dramatic than mass layoffs. Increasingly, it will be attrition + hiring freezes + fewer entry-level positions + one person doing the work previously performed by two or three people. Goldman is already seeing signs of reduced hiring in back-office operations, marketing consulting, graphic design, office administration and call centers.
That distinction is enormously important.
And there’s an intriguing implication of the numbers we’ve now developed across your last three questions: women are ~47% of workers, ~55% of white-collar workers, but perhaps ~58% of AI-displaced workers—and potentially 80%+ of the most economically vulnerable white-collar workers displaced by AI. That is a pretty consequential demographic story that I don’t think has received nearly as much attention as the standard “AI is coming for programmers” narrative.
-
Yeah but women are heavily employed in the state infrastructure. That will not shed jobs because productivity doesn’t correlate with job reduction like it does in the private sector.
The people who will be screwed are the unprotected classes in the private sector white collar cohort. Mostly white males. Probably white males over 50.
be ready.
____________________________________________________________
Sic transit gloria mundi (so shut up and get back to work) -
I think it depends on how it’s presented. The two times in my professional life, I was ordered to lay off people, it was almost exclusively female, because both times I was told to cut headcount…not overall compensation. So I selected the eight assistants, secretaries and coordinators that I could live without, at least over the short run. Both times, it was 100% women, both times HR called me on the carpet, both times, they okayed it.
Why? Because at the end of the day, shit needs to get done, without drama, on time, under budget, and accurately.
-
- You must be logged in to reply to this topic.