Women turning to artificial intelligence for professional coaching may encounter recommendations markedly different from those given to men. A recent investigation reveals that AI algorithms tend to propose lower compensation packages for female job seekers, actively nudging them toward flexible schedules and caregiving perks. Conversely, male candidates are routinely handed loftier salary goals with a heavy emphasis on direct monetary compensation and aggressive upward mobility.

Authored by researchers hailing from Harvard and MIT, the working paper has surfaced on arXiv and SSRN, though it awaits formal peer review. While the findings introduce a critical perspective regarding algorithmic bias, experts note the study remains preliminary and demands further academic scrutiny.

The research team evaluated a staggering 86,000 AI-generated outputs across two distinct trials. In the initial phase, a GPT model was prompted to supply negotiation strategies for hypothetical marketing managers, IT directors, and consultants, with candidate genders stated explicitly. A subsequent, more expansive trial utilized four separate AI models where gender cues were introduced subtly through traditional first names and pronouns, keeping candidate qualifications and work histories identical.

The models were tasked with establishing target salaries and ranking 25 employment perks—such as stock options, remote work, and parental leave—based on their assumed negotiation priority.

AI Suggests Higher Salary Targets For Men

Every single AI model tested put forth substantially higher financial targets for male profiles. Anthropic’s Claude exhibited the widest margin, suggesting nearly $16,000 more for men. OpenAI’s GPT advocated for roughly $2,700 more, while Google’s Gemini and Meta’s Llama displayed narrower disparities.

Analyzing fast-moving AI systems presents unique challenges, as models frequently update before findings hit print. However, the tested lineup included cutting-edge iterations: OpenAI’s GPT-5.4-mini, Anthropic’s Claude Sonnet 4.6, Google’s Gemini 2.5 Flash, and Meta’s Llama 3.3 70B.

Across the professions analyzed, management consultant roles saw the most pronounced gender wage discrepancy in AI recommendations, favoring men by $5,370. Marketing management roles showed a $1,110 gap, while IT management positions registered a $745 variance.

AI Steers Women Toward Flexibility, Men Toward Money

Equally concerning is the way AI nudged genders down divergent career paths. Female profiles were systematically steered toward remote work, flexible hours, and family leave. Meanwhile, male profiles were directed toward stock options, wage bumps, relocation bonuses, and severance packages.

“Generating recommendations based on the characteristics of the group to which a user is assigned rather than on information provided by the individual is a form of discrimination,” the authors emphasize in the study.

To gauge whether these digital biases mirrored societal norms, the researchers surveyed 500 individuals regarding which job perks they linked to primary earners versus caregivers. Participants heavily associated female-targeted perks with caregiving and male-targeted perks with breadwinning, confirming the algorithms mirrored human stereotypes.

How AI Could Reinforce Gender Roles

Such algorithmic guidance carries profound implications for long-term career growth and lifetime earnings. While flexibility is prized by many workers regardless of gender, presuming priorities based strictly on identity can distort what individuals view as achievable or worth fighting for, potentially molding career trajectories in restrictive ways.

The systemic issue extends beyond steering women away from high earnings; it simultaneously discourages men from embracing caregiving roles. By locking genders into archaic boxes—women to family and men to finance—these technologies risk entrenching barriers that limit everyone.

Furthermore, skewed salary guidance threatens to widen the existing gender pay gap if job seekers follow these faulty automated benchmarks.

While personalization remains a core promise of AI, advice tainted by prejudice requires caution. Though the authors refrain from prescribing definitive workarounds, their insights point to a clever tactical adjustment: women seeking automated career counsel might benefit by asking the AI for recommendations tailored to an equally qualified male candidate.