Midlife MavericksYou’re 45. Two kids. A mother who needs you more than she admits. A body that decided to change all the rules at the same time, without warning.
And now they’re telling you you need to learn AI.
Artificial intelligence is not entering a neutral labour market. It’s entering a market that was never built for you. One that measures productivity as if everyone sleeps eight hours and has no one waiting at home.
The International Labour Organization published the numbers. Jobs dominated by women have nearly double the exposure to AI automation compared to those dominated by men. 29% versus 16%. In high-income countries, including Portugal, the gap is even more brutal: 9.6% of female employment is in the maximum-risk category. For men, 3.5%.
The reason is old and nobody wants to say it out loud: occupational segregation. Women are concentrated in administration, HR, customer support, educational coordination, secretarial work. These are exactly the roles that generative AI systems can replicate most easily. CV screening. Scheduling. Data processing. Everything that appears in a job description as “administrative tasks” or “operational support.”
McKinsey estimates that between 40 and 160 million women worldwide will need to change profession by 2030. Not adjust. Change.
Now imagine what this means for a woman of 45 in an office in Lisbon, Porto, Braga.*
(This was written for a Portuguese audience — but you could say the same for London, Birmingham, Manchester. Or New York, Chicago, Boston.*)
Twenty years of career. Manages a team. Runs projects. Does the visible work and the invisible work. Keeps difficult people functioning, absorbs tension, resolves conflicts that never make it to formal meetings. She’s the person everyone turns to when things heat up. And she does all of this while managing a household, school-age kids, a parent with health issues, and a body in perimenopause that has traded her short-term memory and sleep quality. All at the same time. Without a break.
Look at her day. Up at six. Before anything else, she asks the older one if he’s ready for his test. Makes breakfast. Answers three work emails while waiting for the younger one to find his shoes. Arrives at the office feeling like she’s already put in two hours. Meeting at nine. Another at half ten. Some fire to put out in between. By three in the afternoon her brain doesn’t work the way it used to and she knows it, knows that perimenopause has taken that speed she had at 35, but she can’t say this to anyone because nobody talks about this at the office. She tries to leave at five. Around half six she opens the car door to close the work shift and start the dinner, baths, and home shift.
And now they want her, in this space where nothing more fits, to learn how to use prompts, automate workflows, master tools that change version every two weeks.
Reports that recommend continuous training, digital upskilling, and lifelong learning start from an assumption they rarely say out loud: that the person has time. That the person gets home and has two free hours for an online course. That person doesn’t have a ten-year-old, or a seventy-five-year-old mother, or night sweats that wake her at three in the morning.
I know this story. I’ve heard it more than once. Eighteen, twenty years in coordination or HR roles. One day they install a new system. It screens CVs, schedules interviews, generates performance reports. And nobody asks what the person was doing beyond that.
I know what she was doing beyond that. She spent hours talking to rejected candidates so they’d leave the process with their dignity intact. She managed the fragile egos of directors who didn’t know how to give feedback. She resolved tensions between departments that nobody wanted to acknowledge. She was the emotional glue holding an entire organisation together. But none of this appears in a job description. None of this appears on a dashboard.
Bloor Research found that around 90% of job descriptions don’t reflect what people actually do. AI is being trained on formal representations of work. When a company feeds its job architecture into an automation system, it’s automating what someone wrote down once, not what people actually do for eight hours. The invisible work disappears. And the people who did it disappear with it.
But I refuse the victim narrative. I refuse it with everything I have.
Because telling a woman of 45, overloaded, in perimenopause, with a care burden that never stops, that AI is an “opportunity” without first acknowledging that the system she works in was built to disadvantage her, is insulting. It’s like telling someone who’s drowning that the solution is to swim faster.
And at the same time, that woman is not fragile.
Twenty years managing egos, absorbing tension, making broken things work. Reading a room and knowing who needs what before anyone opens their mouth. No algorithm replicates this.
The World Economic Forum has already warned: AI risks deepening existing gender inequalities. The jobs that will disappear are theirs. The ones that will be created, not so much. And 44% of commercial AI systems already show measurable gender bias. The systems deciding who gets hired and who gets let go were built mostly by men and trained on historical data reflecting decades of discrimination.
Give me a break.
The question nobody asks is this: who’s sitting at the table when they decide what to automate?
If women aren’t in that room, if midlife women aren’t in that conversation, their jobs will be eliminated.
No algorithm knows how to do what you do. The only question is whether you’re going to wait for a system that never valued you to decide to protect you. Or whether you’re going to take your twenty years and build something else.
Twenty years of experience don’t disappear because a robot screens CVs.
Use them. They’re yours.
Comment below. What work do you do that has never appeared in any job description? I read every response.
This article was first published in Portuguese in my weekly column Oh pá, não me lixem! for Executiva.
