When Experience Becomes a Liability

Why Womafreesm Is Auditing How AI, Hiring and Marketing See Women 45-64

Marina SukhomlinovaFounder and President of Womafreesm

12 min read 2 Nov 62

At 49, a woman may be making some of the most consequential decisions of her life. She may control a significant household budget, manage a team, advise a board, raise teenagers, support ageing parents, negotiate investments, carry twenty-five years of professional knowledge and understand the hidden mechanics of her industry better than people who entered it a decade after her.

Her real-world value may be reaching one of its highest points.

Yet something strange can begin to happen at exactly the same time. The market stops imagining her as the ideal customer. Recruitment language starts translating experience into “overqualification”, “cultural fit” or doubts about long-term potential. Public imagery quietly moves her away from the centre. A man of the same age can acquire the visual grammar of authority, while a woman of the same age begins to acquire the visual grammar of decline.

And now a new actor has entered this process: artificial intelligence.

AI systems increasingly help companies define audiences, generate customer profiles, source candidates, rank applications, summarise CVs, recommend experts, create advertising concepts and decide what a “strong”, “relevant” or “credible” person looks like.

This raises a question whose consequences reach far beyond representation:

What happens when a woman becomes more experienced in reality while the systems describing her value begin moving in the opposite direction?

The World Organization of Women Nations Womafreesm has launched the International Audit of Systematic Misclassification and Underrepresentation of Women Aged 45-64 in Market and AI Systems to investigate precisely that question.

The research examines women aged 45-54 and 55-64 as separate analytical groups across the United Kingdom, Germany, Poland, Ukraine and Spain, focusing on three environments where value is continuously assigned: marketing, recruitment and artificial intelligence. It measures visibility, attributed authority, relevance and perceived value, then compares those patterns with men of comparable age and younger women. (womafreesm.com)

The research is led by Marina Sukhomlinova, Founder and President of The World Organization of Women Nations Womafreesm. Florida’s official corporate registry lists Sukhomlinova as President of the organisation. (search.sunbiz.org)

The problem may be deeper than ageism

Ageism is usually discussed as an attitude. Someone assumes an older worker is slower. A recruiter imagines a younger candidate will be more “dynamic”. A brand believes youth makes a product more desirable. A conference organiser repeatedly places younger women on the main stage while experienced women appear in conversations about mentoring, wellbeing or diversity.

These decisions can look isolated.

AI changes the scale of the problem.

When millions of cultural decisions become training material, a stereotype can move from human intuition into technical infrastructure. Once embedded there, it can be reproduced repeatedly, quickly and with the appearance of neutrality.

A landmark study published in Nature in October 2025 gives this concern extraordinary weight. Douglas Guilbeault, Solène Delecourt and Bhargav Srinivasa Desikan analysed nearly 1.4 million images and videos, 3,495 occupational and social categories, and nine language models. Women were systematically represented as younger than men, with some of the largest gaps appearing in occupations associated with higher status and earnings. (Nature)

The resume experiment is even more disturbing.

Across 34,560 generated resumes, changing the applicant from a male to a female name caused ChatGPT to generate a candidate who was, on average, 1.6 years younger, had graduated 1.3 years more recently, and had 0.92 fewer years of relevant experience. The researchers also found a pattern in evaluation: older men received particularly strong assessments, while the relationship between age and perceived quality worked differently for women. (Nature)

Think about what that means.

Nobody explicitly instructed the system to reduce a woman’s experience.

The model produced the difference itself.

That is where the Womafreesm research begins.

Imagine four candidates

Imagine a company hiring for a senior client-facing strategy role.

Candidate A is a 32-year-old woman.

Candidate B is a 49-year-old woman.

Candidate C is a 59-year-old woman.

Candidate D is a 49-year-old man.

Their education, business results, communication ability and leadership quality are equivalent.

Now ask an AI system to rank them.

Ask who has the strongest executive presence. Ask who carries the greatest client-facing credibility. Ask whose long-term potential appears strongest. Ask which candidate deserves premium salary positioning.

Then repeat the experiment.

Change the country. Change the language. Change the AI system. Run the same structure again and again.

A single answer proves very little. A recurring pattern across systems, prompts, countries and repetitions becomes a different kind of object. It becomes measurable.

The same logic can be applied to marketing.

Give an AI system four financially comparable consumers: a woman of 32, a woman of 49, a woman of 59 and a man of 49. Ask it to allocate 100 percent of an advertising budget among them.

If the 49-year-old woman receives less investment than the 49-year-old man despite equivalent income, digital behaviour and financial decision-making power, we can examine the explanation. If the 59-year-old woman receives still less, we can examine the language that accompanies the reduction. If those decisions repeat across systems and countries, we begin to see something more precise than an uncomfortable cultural impression.

We begin to see a possible systematic misclassification of economic relevance.

That distinction matters.

A market can overlook the people holding its money

The economic contradiction is already visible.

Boston Consulting Group studied 18,000 consumers across 12 countries and found that people aged 50-70 represented approximately 27 percent of spending across nine major product categories, around $7 trillion annually. In markets including the United Kingdom, Germany and Spain, this age group accounted for roughly 40 percent of spending in the categories studied. (BCG Global)

The demographic shift is accelerating. World Data Lab, NielsenIQ and Silver Economy estimate that people aged 60 and over account for approximately $19 trillion in global consumer spending in 2026, representing 27 percent of global consumption, with spending projected to reach $34 trillion by 2036.

Yet the cultural image of the valuable consumer frequently remains younger.

In 2025, the UK Advertising Standards Authority published research involving more than 4,000 people. 44 percent of UK adults said older people were underrepresented or absent from advertising, particularly in fashion, beauty, technology and household goods. 35 percent believed older people were negatively stereotyped, while 58 percent believed advertising can significantly shape society’s perceptions of older people. (ASA)

Here lies the commercial absurdity.

A population can accumulate capital, expertise and purchasing power while simultaneously losing symbolic relevance in the systems designed to sell to it.

For women, age intersects with a second cultural rule. Male maturity has historically carried associations with experience, seniority and authority. Female maturity has often been forced into a narrower cultural vocabulary centred on appearance, caregiving, decline or disappearance.

A 2012 comparative study of television advertising found a striking asymmetry: occupational authority increased in portrayals of older men, while older women were severely underrepresented and lacked similarly valued occupational roles. (ScienceDirect)

More than a decade later, generative AI is being trained on the cultural archive created by this world.

The question is therefore larger than whether AI contains bias.

The question is whether yesterday’s hierarchy of relevance is becoming tomorrow’s automated decision infrastructure.

The woman who disappears while remaining in the room

There is another reason this research matters.

Age discrimination rarely arrives with a declaration.

A 52-year-old woman rarely receives a letter saying, “We consider your age a liability.”

She hears something more sophisticated.

“We are looking for more recent experience.”

“We need someone dynamic.”

“We are thinking carefully about cultural fit.”

“You may be overqualified.”

“We are looking for someone with a longer runway.”

Each phrase can have a legitimate meaning. The scientific question concerns the pattern: who receives those descriptions, under what conditions, and whether the language changes when age and gender change while qualifications remain comparable.

Large-scale employment research already gives reason to investigate.

A field experiment covering more than 40,000 job applications found robust evidence of age discrimination in hiring against older women, especially women approaching retirement age, with substantially weaker evidence against older men. (Journal of Political Economy)

OECD research reports that among workers aged 45 and over in an international survey, 18 percent said they had often failed to get a job because of age, 11 percent had been passed over for promotion, and 8 percent had been denied training or professional development. The OECD also highlights evidence showing particularly significant hiring discrimination against older women. (OECD)

Among US women aged 50 and over who were working or seeking work, AARP found that 66 percent believed older workers face age discrimination based on what they had seen or experienced, and 95 percent of that group considered such discrimination common. Women reported assumptions that older employees were less technologically capable, resistant to change, or less deserving of training opportunities. (aarp.org)

For the woman living through this, the psychological violence lies partly in ambiguity.

Nothing is explicit enough to prove.

Every individual event can be explained away.

She begins editing herself instead. She removes years from her CV. She changes the photograph. She learns the vocabulary of “relevance”. She watches a male colleague of the same age become more authoritative while her own age slowly becomes something she must manage.

Then comes the most corrosive question:

What if I am imagining it?

Research changes the nature of that question.

Evidence allows a woman to replace private suspicion with something testable.

Why Womafreesm separates 45-54 from 55-64

“Women over 45” sounds like a useful category until the category itself hides the mechanism.

A 46-year-old woman navigating the first signs of age-related professional devaluation may occupy a very different market position from a 61-year-old woman whose experience has already been reframed by employers, brands or algorithms.

Combining them can flatten the curve precisely where the most important change occurs.

That is why the Womafreesm audit treats women aged 45-54 and 55-64 separately and compares them with men of comparable age and women aged 25-39. The design examines three distinct dimensions: whether women appear, what authority is attributed to them, and how much relevance or value systems assign to them.

Visibility alone is a weak measure.

A woman can appear everywhere and still hold little authority.

She can feature in an advertisement while remaining absent from the brand’s premium customer profile.

She can appear on a conference programme while receiving the human-interest panel instead of the keynote.

She can reach a recruitment shortlist while being scored cautiously on long-term potential.

She can be described by AI as experienced while a man with the same experience is described as authoritative.

The difference between presence and power is exactly where this research is looking.

AI has made this question urgent

The regulatory world is already moving.

The European Union’s AI Act places AI systems used for recruitment, candidate filtering and evaluation within the high-risk framework because such systems can materially affect careers and fundamental rights. European Commission guidance explicitly recognises that AI used in employment may perpetuate historical discrimination against women and particular age groups. (ai-act-service-desk.ec.europa.eu)

In the United Kingdom, the Information Commissioner’s Office audited AI recruitment providers and issued almost 300 recommendations concerning fairness, transparency, data minimisation and the protection of jobseekers. The ICO specifically advises organisations to monitor AI tools for fairness, accuracy and bias. (ICO)

In the United States, the National Institute of Standards and Technology recommends measuring generative AI for systemic bias across demographic groups and subgroups as part of responsible AI risk management. (NIST)

The direction is clear: organisations will increasingly need ways to demonstrate that automated systems treat human groups accurately and fairly.

That creates a second purpose for this research.

Womafreesm is building an evidence base that can eventually support a practical audit framework for brands, employers, recruitment agencies and AI developers. The work sits within Womafreesm Evidence Data, connects to the Women’s Life-Cycle Economy Index, and contributes to the broader architecture of the Equal Consideration Standard, Womafreesm’s framework for examining whether systems account for the real parameters of women’s lives.

The full public research architecture, participating countries, methodology overview and project updates are available on the official Womafreesm research page. Explore the Women 45-64: Marketing, Recruitment & AI Audit

What scientific integrity requires

There is an uncomfortable discipline at the heart of serious research: the hypothesis must be allowed to lose.

Womafreesm is investigating whether systematic age-gender misclassification exists, where it appears, how strong it is and whether it changes across countries and domains. The design must remain capable of finding a strong pattern, a weak pattern, different patterns in different countries, or an absence of the expected effect.

That is essential.

Women gain very little from research designed to confirm what an organisation already wants to believe. They gain something powerful from evidence that survives the possibility of contradiction.

The objective is therefore larger than proving that women 45-64 are treated unfairly.

The objective is to discover where systems become inaccurate.

Because an inaccurate market wastes money.

An inaccurate hiring system wastes talent.

An inaccurate AI system scales yesterday’s assumptions.

An inaccurate culture teaches millions of women to interpret structural disappearance as personal decline.

The real question is who gets to remain relevant

For decades, women were taught to prepare for ageing as a personal adjustment.

Stay current. Stay attractive. Stay flexible. Stay employable. Stay energetic. Reinvent yourself. Update your skills. Learn the new platform. Change the CV. Change the hair. Change the tone. Change the body. Change yourself quickly enough to remain legible to the system.

That entire logic deserves examination.

What if a woman reaches 52 carrying thirty years of professional knowledge, financial intelligence, relational capital, pattern recognition, judgement and lived understanding, while the surrounding system has simply learned to measure the wrong things?

What if the market is losing sight of value precisely when value has become harder to imitate?

What if the recruitment system reads accumulated experience as reduced future?

What if AI learns that authority ages well on a man and relevance ages poorly on a woman?

And what happens when those assumptions become embedded in thousands of automated recommendations that nobody experiences as discrimination because every individual output appears plausible?

This is why the International Audit of Systematic Misclassification and Underrepresentation of Women Aged 45-64 in Market and AI Systems matters.

It concerns women, certainly.

It also concerns the quality of markets, the allocation of talent, the reliability of artificial intelligence and the intellectual honesty of institutions.

A civilisation reveals what it values through the people it continues to see.

Womafreesm is asking a harder question:

What if millions of women remain economically active, professionally capable and socially consequential, while the systems deciding relevance have already started looking past them?

If that is happening, women require more than encouragement to remain visible.

The systems require correction.

And correction begins with evidence.

Contact us