AI Skills and the Gender Gap: Who Gets Ahead in the Changing Job Market? | HackerNoon

This post was originally published on this site.

I should have been born into a wealthy family. Preferably a boy, and ideally in the US. This article is my attempt to put a few numbers behind that joke. The numbers, unfortunately, are not so funny.

The World Economic Forum projects that AI and information-processing technologies could create 11 million jobs on the one hand and displace 9 million on the other by 2030, resulting in a net gain of around 2 million [1]. Used responsibly, AI could improve productivity and make some jobs better. Encouraging on paper, isn’t it?

But who influences these shifts, gets the opportunities, tools, training and freedom to benefit, and who is left adapting to decisions made elsewhere? Spoiler: the starting point is far from equal.

Round one: Who gets to train?

Numerous research studies reviewed in this article point to several overlapping circumstances that tend to disadvantage particular groups.

The most concerning patterns are as follows:

  • Women face greater exposure to disruption in some types of work while remaining underrepresented in AI careers.
  • People with higher levels of formal education are also more likely to receive AI training and use the technology. But if AI training mainly reaches graduates already working in well-supported roles, are we giving the most support to those who already have the strongest starting position?
  • Age matters too, although its influence is intertwined with occupation, education and access.
AI training: use and roles
AI training: age groupsAI training: age groups

Since these inequalities existed long before AI, the question is whether AI helps close them or simply gives them a faster, more efficient way to persist.

Let’s look at what is happening across different job markets, who is getting ahead, and who may be expected to keep up without being given much help.

Same game, different markets

Demand for AI skills varies considerably by region and occupation. In Europe and the US, most of this demand is concentrated in a handful of occupational groups. In Latin America, it is spread across a wider range of jobs [2].

AI skills: occupation demand across regionsAI skills: occupation demand across regions

Computing leads, but creative work is catching the eye

  • In Europe, Computer and Mathematical occupations lead by a wide margin: 60% of workers in these occupations are in roles requiring AI skills. No other occupational group in the region exceeds 13%. Computing also leads in the US, at 44%, while Latin America reaches 34% [3].
  • Latin America has another occupational group at the same level: Arts, Design, Entertainment, Sports and Media, also at 34%. The figures are much lower in Europe, at 5%, and in the US, at 7% [3]. This is perhaps the most interesting contrast. We often think of AI skills as belonging to developers and data specialists. But in Latin America, creative and media work is just as prominent by this measure.
  • In the US, Management ranks second at 19%, followed by Legal at 14%. The Legal category reaches 10% in Latin America and 5% in Europe [3]. Perhaps AI is becoming part of the management toolkit too, something managers increasingly need to understand before deciding what to delegate, automate or leave to people.
  • The HKU–Deloitte survey covering China and Hong Kong offers a different perspective: Customer Service leads at 58%, followed by Marketing at 54% and IT/Technology at 53%. Overall, AI adoption is distributed quite evenly across industries, with average usage exceeding 40%. This makes the Chinese market somewhat unusual compared with the others. However, these differences could also be explained by variations in research methods and methodology [4].

Count the workers, and the picture changes

Looking at the number of workers, Europe and the US show a similar pattern: three occupational groups account for around 75% of AI-skills demand – Computer and Mathematical occupations, Management, and Business and Financial Operations. Latin America’s demand is less concentrated, with seven occupational groups accounting for around 80% [3].

Having said that, could it be that Europe and the US are focusing on building and directing AI, while Latin America is focusing on putting it to work more widely? China seems to sit somewhere in between, apparently doing both.

Round two: What you do may determine how long you stay in the game.

Women and men have historically been concentrated in different types of work. Women are more heavily represented in healthcare, teaching, social work and business administration, while men dominate construction, manufacturing, skilled trades, science and engineering, IT, protective services, transport and agriculture.

In addition, women are disproportionately concentrated in routine and repetitive jobs, particularly clerical, administrative and business support roles. These jobs often involve tasks that generative AI can handle, potentially leaving women more exposed to changes in their work and, in some cases, job displacement.

ILO research published in March 2026, covering data from 84 countries, shows that so-called “female-dominated occupations” are almost twice as likely to be exposed to generative AI as male-dominated ones: 29% versus 16% [5]. A World Bank and ILO study of Latin America and the Caribbean found that 2–5% of jobs face potential full automation, with women facing around twice the risk of men on average [6].

Meanwhile, access to AI careers is not becoming more equal everywhere. In the UK, women’s share of AI roles fell to 20% in 2025, down four percentage points since 2020 [7]. This measures something different, but the combination is uncomfortable: greater exposure to disruption in existing jobs, alongside limited access to the careers shaping the technology.

And even in industries where women make up most of the workforce, they do not necessarily use AI as much as men. Canadian research found that:

  • In healthcare and social assistance, 21% of men reported using generative AI at work, compared with 13% of women.
  • In professional, scientific and technical services, where men make up a larger share of the workforce, the figures were 55% for men and 47% for women [8].

Different countries, different labour markets, different research teams. Yet the pattern keeps repeating: women are more exposed to AI-driven disruption, but have fewer opportunities to shape the technology or benefit from its adoption.

Round three: Who is building the game?

Women remain underrepresented in AI, not only as users in some types of work, but also as developers, creators and decision-makers. This limits their influence over how the technology is designed and deployed, as well as the range of perspectives and experiences brought into the process [9].

According to ILIA 2025, across the countries covered, t**he concentration of AI engineering talent among men is, on average, 2.5 times higher than among women** [10]. The gap in AI literacy skills is smaller, although it still favours men. So, while the adoption of general-purpose AI tools is more gender-balanced, building the technology remains a much more heavily male-dominated domain.

Unequal access shapes who builds AI, while biased data and design choices can shape how AI treats people. And the consequences can extend far beyond who gets the job:

“AI systems and their “black box” algorithms may pose significant challenges for gender equality. Gender bias embedded in AI systems may lead to unequal allocation of resources, information, and opportunities. Because these systems are trained on historical data, pre-existing gender inequalities may be learned and reproduced by algorithms, placing women at a disadvantage in career promotion and thereby reinforcing gender inequality in the labour market.”[11]

Part of this imbalance likely starts with women’s underrepresentation in STEM (science, technology, engineering and mathematics), which provides important pathways into AI engineering. This, in turn, widens the gender gap in AI-related fields [10]. And these differences begin to emerge long before the first job application:

“Gender differences in interests, aptitudes and aspirations widen with age, suggesting that, at some point, girls are discouraged from scientific careers. At age 15, boys and girls perform similarly in science and mathematics tests, however, only 1% of girls report that they want to work in ICT-related occupations, compared to 8% of boys. The lack of female role models in STEM has been shown to have an impact on girls’ subject choices in school and university. Additionally, discrimination in recruitment processes within technology and STEM fields results in women being less likely to gain employment, even when they possess the necessary skills.”[12]

Sue Duke of LinkedIn warns that, at the current pace, girls born today would not see equal representation at the top within their lifetime, with the gap even more pronounced in AI roles [13].

Not exactly the future-of-work promise we were hoping for…

It seems women are caught in a double disadvantage: they remain underrepresented in the development of AI, leaving fewer opportunities to influence technology that may reproduce existing biases and discrimination. At the same time, they may face greater consequences from the technology’s development and adoption in the labour market. In other words, women have less say in building the technology, yet may have more to lose from the way it changes the world of work.

But then, what does the job market look like now?

Women are consistently less represented in AI firms than in comparable non-AI firms, even when comparing employees at the same seniority levels [9].

At the individual-contributor level, women account for 36.6% of employees in AI firms, compared with 44.8% in non-AI firms. At the top-management level, the figures are 25.3% and 27.7%, respectively [14].

Hiring tells a similar story. LinkedIn’s analysis of the US labour market found that women accounted for just 26% of new hires in AI roles in 2025, compared with 50% in non-AI occupations [15].

Women have a stronger presence in lower-paid AI roles, such as data annotation, where they account for just over half of new hires – one of the few AI-related roles with a relatively balanced gender split. Meanwhile, AI roles attracting more male hires tend to offer higher advertised salaries.

So, the gap is not just about who gets into AI. It is also about who gets access to its best-paid opportunities and who gets to shape what the technology does next. But getting into the game is one thing. Having the resources, experience and opportunities to stay in it is another.

AI jobs: roles. leadership and paychecksAI jobs: roles. leadership and paychecks

Season one finale: Not a fan of these statistics. Are you?

AI may create more jobs than it removes. But a positive net figure does not tell us who gets the new opportunities. The numbers do not point to one inevitable future (although the pessimistic part of me, which I prefer to call “rational”, might argue otherwise). They do, however, highlight different starting positions: who gets training, who enters the better-paid roles, and who has a say in how AI changes the way we work.

The gap between educated people with access to technology and those with less education is widening. Gender inequality is making a comeback, with renewed force, as AI is trained on historical data that is largely biased and prejudiced. After 45, does the door to AI training start to close? At least, that is how I read the age-pattern graph. And only strong economies are competing to develop these tools, while others are left merely using them.

Having worked in IT for almost my entire career, I am in a relatively privileged position. My employer invests in training, including AI training, and gives us access to tools we can actually practise with. Learning is part of the job, rather than another task to squeeze into the evening.

And yet, I recognise myself in some of these patterns. I use AI, but I am not building its models or deciding what goes into its “black box”. And, confession time: I mostly use it for data analysis and research. One more woman adding a data point to the charts in this article.

I could give my younger self some advice, I would encourage her to explore STEM and build more confidence in technical subjects. I would still keep the humanities; they are probably responsible for half the questions I am asking here. But I would want more freedom to move between understanding technology and helping to build it.


  1. World Economic Forum, Future of Jobs Report 2025: Jobs Outlook.
  2. A small but important distinction: McKinsey’s figures come from requirements in job advertisements. They indicate demand for AI skills, rather than directly measuring how many workers actually use AI.
  3. McKinsey Global Institute, Agents, robots, and us: United States (November 2025); Europe (May 2026); Latin America (July 2026).
  4. HKU and Deloitte, China AI Adoption Index 2026.
  5. International Labour Organization, Gen AI, occupational segregation and gender equality in the world of work.
  6. World Bank Group and ILO, Buffer or Bottleneck? Employment Exposure to Generative AI and the Digital Divide in Latin America (2024).
  7. UK Department for Science, Innovation and Technology, AI Labour Market Survey 2025: Executive Summary.
  8. Statistics Canada, Workplace artificial intelligence use: A profile of sociodemographic and job characteristics.
  9. UNESCO, Gender Equality in AI.
  10. ECLAC/CENIA, Latin American Artificial Intelligence Index (ILIA) 2025: Key Findings, Applied Artificial Intelligence, and Human Talent, “A Persistent Gender Gap” section, page 90.
  11. ILO, Artificial intelligence application and gender wage inequality: evidence from China (2026)
  12. OECD/Korea Labor Institute, Artificial Intelligence and the Labour Market in Korea.
  13. World Economic Forum, Twenty Years of Progress on Gender Parity Now Fragile, WEF Report Finds
  14. World Economic Forum, Global Gender Gap Report 2026: Trends and Pathways Shaping the Future of Parity.
  15. LinkedIn Economic Graph Research Institute: The Triple Penalty: Mapping the Gender Gap in the AI Economy & The AI Talent Divide: Strong Demand and Pay, Uneven Returns, 2026.

Leave a Reply

Your email address will not be published. Required fields are marked *