Four Months on Leave, and the AI Models Changed

We Thought a Young Industry Would Be Fair

There has long been an assumption in the AI industry: because no one can have 10 years of experience with a model released just 1 year ago, everyone starts from roughly the same line. If career length is no longer an advantage, the thinking went, entrenched disparities from older industries wouldn’t take root here.

Urvashi Batra, co-founder and CEO of Prioriwise, an AI platform for IT services, disagrees. In a recent media interview, she pushed back against the claim that AI levels the playing field for women by lowering barriers to entry. As an AI founder, she felt taken less seriously than her male co-founder. The pair learned firsthand that investor pitches had a noticeably higher chance of securing funding when he took the lead.

In the same report, Jayeeta Putatunda, an AI engineering lead at Turing, shared an even more concrete account. When she returned from a 4-month maternity leave, the frameworks had shifted entirely, and the models were on another level. Catching up was daunting; she credited supportive colleagues and a husband who split childcare equally for making it possible.

Today, let’s examine these two stories alongside recent figures from LinkedIn. In an industry where a fresh slate was supposed to guarantee fairness, rapid iteration cycles are instead turning time away into a steep penalty.

The Higher the Pay, the Lower the Share

The underlying figures come from two reports published in July and August 2026 by the LinkedIn Economic Graph team. The Triple Penalty analyzed tens of millions of employees at mid- to large-sized enterprises across 27 countries, while The AI Talent Divide examined U.S. job postings and hiring data from 2023 to 2026.

In the U.S., women accounted for 26% of new hires in AI roles¹ in 2025. In non-AI roles, that figure was 50%. Over the same period, AI job postings doubled compared to 2023, and the median compensation offered in a typical AI posting was roughly $177,000—more than twice that of non-AI postings (around $80,000). In other words, in the fastest-growing and highest-paying positions, women hold only about half the representation they do elsewhere.

Breaking the numbers down by role makes it clear that this gap does not emerge at a single point.

Role	Share of Women Among New Hires	Disclosed Compensation
Data annotator2	Roughly half	Median ~$51,000
Director of AI	26%	Unconfirmed
Head of AI	20%	Typically ~$236,000
Member of Technical Staff3	18%	Unconfirmed

LinkedIn noted that women accounted for more than half of new hires in AI roles with lower educational barriers and a high prevalence of hourly, project-based, or remote work. Conversely, in top-paying positions like Head of AI, Director of AI, and Member of Technical Staff—where over 95% of workers hold at least a bachelor’s degree—the female share drops to around 20%. Rather than a single wall blocking entry at the gate, it looks more like a slope that tilts steadily steeper as compensation rises.

The article that prompted this piece reported that women working in AI are concentrated in low-paying roles like data annotator. According to the LinkedIn figures reported by Axios, the gender ratio among annotators is roughly 50-50. These two statements can both be true because their denominators differ: the former looks at which roles account for the largest share of women working in AI, while the latter measures the gender breakdown within a single role. The same article reported that across all AI roles, the median salary for men was $45,000 higher than for women, though I could not verify this figure directly in LinkedIn’s original report.

The global numbers point in the same direction. Across AI companies in 27 countries, women make up 13% of C-suite executives overseeing AI. Looking at all director-level and above roles, women account for 38.5%, but narrowing the scope to AI leadership positions drops that figure to 31.2%. Even after controlling for country and role differences, LinkedIn found that the female share was about 15 percentage points lower at the executive level, about 10 percentage points lower in AI roles, and about 5 percentage points lower at AI companies—a phenomenon it dubbed the “triple penalty.”

A few caveats are worth keeping in mind. The data is based on profiles and job postings on LinkedIn’s platform, and as Axios pointed out, AI jobs still represent a tiny slice of the overall labor market. Yet it is the fastest-rising slice in terms of pay, meaning disparities opening up here could feed into broader average wage gaps over time.

Why a Young Field Became an Even Narrower Door

In a field where track records are uniformly short, standard resume credentials lose their sorting power. When nobody has 10 years of experience, hiring managers search for substitute signals—and that void is typically filled by referrals, personal connections, and familiar default filters. Brenda Darden Wilkerson, president of the women-in-tech advocacy group AnitaB.org, notes that while companies are hiring at breakneck speed, they continue to rely on the same networks and referral pipelines, which have never offered women equal exposure. Batra’s experience in pitch meetings reflects the exact same dynamic: the weaker the objective basis for judgment, the more easily the identity of the speaker becomes the evidence.

The other factor is velocity. Putatunda observed that anyone harboring big ambitions in this field eventually faces periods that demand working well past regular hours, where 12-hour workdays are far from uncommon. Nor is there always a shortcut around putting in that time. She summed up why women are being pushed out of AI like this: it is not a lack of interest or capability, but rather the absence of the underlying support system needed to keep pace.

Korean already has a phrase for this situation: gyeongnyeok danjeol, or career disruption. In other industries, career disruption was a penalty accrued over years. But in an environment where tools and models churn every few months, a far shorter gap can produce a similar outcome. In Putatunda’s case, that period was 4 months.

There is no need to pick between the two explanations. Because evaluation criteria are thin, companies hire who they know; because change is rapid, anyone who steps away falls behind easily. Both forces act together against those far more likely to step away for childbirth and caregiving.

The institutional supports are receding

Historically, the tech industry’s go-to answers for the gender gap have been mentorship and diversity targets. The results have been mixed. According to the original reporting, women’s share of US tech jobs has climbed to roughly 1/3, but women still leave the industry at far higher rates than men. Over the past few years, however, many corporations have scaled back their DEI⁴ programs.

Felicia Newhouse, founder of AI Powered Women—an organization dedicated to expanding women’s participation in AI—pointed to the US Department of Justice scrutinizing DEI initiatives under anti-discrimination laws. Gaining corporate access for organizations with “women” in their name is becoming increasingly difficult, she notes, and internal allies are finding it harder to sustain programs aimed at women. Her deepest concern is that a gap in participation is hardening into a gap in power.

This is precisely why that earlier 13% figure carries so much weight. The C-suite decides where to introduce AI across a company’s operations, what metrics to evaluate it by, and whose work to reshape. LinkedIn’s research showed that women are disproportionately concentrated in functions most vulnerable to AI transformation, such as customer service. The sharp disparity in gender ratios between those whose work will change and those deciding how it changes deserves scrutiny in its own right.

Oswarld’s Lens

When designing enterprise deliveries or consulting engagements, I always work on the premise that “models are constantly improving.” The more this premise holds, the more working in this industry demands constant, uninterrupted presence. If the system leaves individuals to bridge whatever shifted during a few months away entirely on their own, that cost is billed first to those with the most compelling reasons to step away.

That is why what worries me most about these figures is the moment a structural disparity gets misread as a difference in competence. When the speed of catching up becomes the primary signal for hiring and promotion, someone who was away for 4 months is judged by their gap before their ability. That is also why Putatunda’s choice of the word “foundation”—rather than will or passion—resonated so strongly with me.

Closing

Claiming that an industry offers a level playing field simply because it is so young remains wishful thinking. Youth alone has not thrown the doors wide open, and the numbers show that those doors only narrow as compensation climbs higher.

---

💬 In your field, Reader, how long does it take to catch up after taking a few months off?

📨 If you know a colleague preparing to return from leave or someone looking to transition into AI, please share this issue with them.

---

References & Further Reading

• Women are missing out on the AI jobs boom · Axios, 2026-08-18. Covers US AI new hire gender ratio (26:74), data annotator gender ratio, and median compensation.
• Women are being left behind in high-paying AI jobs boom, LinkedIn study finds · CBS News. Covers the US 26% vs. 50% split and 13% female AI corporate leadership across 27 countries.
• Women hold just 13% of AI leadership roles despite salaries more than doubling · Mediaweek. Details titles and scopes of both LinkedIn reports, director-level and above representation (38.5% vs. 31.2%), and hiring share by role.
• Women hold just 1 in 4 AI roles, LinkedIn finds · HR Brew. Breaks down the three stages of the “triple penalty” and the compensation gap ($177,000 for AI postings vs. $80,000 overall).
• Fortune coverage of LinkedIn’s “Triple Penalty” report · Fortune, 2026-08-31. Reports on typical Head of AI compensation reaching $236,000.
• AI jobs are booming and paying more, but women are being left behind · FOX. Discusses the bachelor’s degree-or-higher requirement rate in the highest-paying roles.

The author is Oswarld (Kwangseob Ahn). Current roles: Adjunct Professor at Sejong University, Strategy Consultant at INLEVEL9. Career, research, books, and recent work are kept current on the About page. Latest · July 2026: HEMA-2: A Consolidation-Aware Tri-Memory Architecture with Multi-Channel Scheduling for Lifelong Conversational AI.

📝 Glossary

Footnotes

1. AI roles: Roles categorized by LinkedIn as requiring AI skills as a core competency, including AI Engineer, Machine Learning Engineer, and Head of AI. ↩

2. Data Annotator: Workers who label training data for AI models or review model outputs, predominantly employed on an hourly or project-based contract basis. ↩

3. Member of Technical Staff (MTS): A common title at AI research labs describing technical roles that bridge cutting-edge research and hands-on product development. ↩

4. DEI: Diversity, Equity, and Inclusion; corporate initiatives and frameworks aimed at increasing the representation of underrepresented groups in hiring and promotion. ↩