AI negotiation advice steers women toward caregiving benefits: study

Artificial intelligence (AI) is reshaping work in ways that are not gender-neutral. 

A March 2026 brief from the International Labour Organization (ILO) found that 29 per cent of female-dominated occupations are exposed to generative AI, compared with 16 per cent of male-dominated ones.

And new research suggests AI negotiation advice may also be steering careers, telling women to ask for less money and prioritize caregiving benefits while pushing men toward equity and advancement – with implications for employers' pay equity and human resources (HR) practices.

The study – posted to the arXiv repository on Oct. 1, 2026 –  analyzed 86,000 queries to four large language models (LLMs): OpenAI's GPT, Anthropic's Claude, Google's Gemini and Meta's Llama. Its authors are Nadja Born and Hannah Riley Bowles of Harvard Kennedy School and Jared R. Curhan of the MIT Sloan School of Management, all in Cambridge, Mass. The paper has not been peer reviewed.

"The same systems that expand access to negotiation coaching may simultaneously reproduce societal expectations about who should prioritize advancement and paid work versus who should prioritize accommodation and unpaid work," they wrote.

Lower salary targets, different negotiation agendas

Every model recommended higher salary targets for men than for otherwise identical women. Across the four models, the gaps ranged from US1,594forLlama3.370BtoUS15,915 for Claude Sonnet 4.6.

The gaps went beyond salary. In a first experiment using a GPT model, 80 per cent of mentions of childcare support were aimed at women, as were 69 per cent of mentions of parental leave. Ranking 25 negotiable terms, the four models favoured remote work and parental leave for women, and equity, relocation and severance for men.

Forbes contributor Kim Elsesser noted the stereotyping cuts both ways, also steering men away from caregiving. The researchers did not test whether users act on the advice.

Table 1: AI negotiation advice findings, from Born, Curhan and Bowles, arXiv, Oct. 1, 2026

Why non-salary terms escape pay equity reviews

The authors see greater risk in non-salary terms. "Explicit salary disparities tend to be more visible and therefore amenable to auditing, monitoring, and interventions than non-salary employment terms, which are often more discretionary, individualized, and opaque," they wrote.

Canadian pay transparency rules sharpen that contrast. Since Jan. 1, 2026, Ontario employers with 25 or more employees must post expected pay in public job ads, as detailed in our guide to Ontario's new hiring transparency requirements. By June 2026, 71 per cent of Ontario postings included pay information, according to Indeed data cited in our analysis of how many Canadian job postings include salary information.

The federal Pay Equity Act requires federally regulated employers with 10 or more employees to proactively review compensation through a pay equity plan. Neither regime reaches the chatbot a candidate consults before an interview.

What AI negotiation advice means for HR leaders

That leaves employers best placed to spot the pattern. Employers hold the data needed to compare the terms offered to men and women, the authors noted, while candidates cannot see the advice an identical counterpart received.

For HR leaders, that points to pay equity reviews of the full offer package – equity, relocation and flexibility – rather than base salary alone, along with scrutiny of employers' own AI tools. Calgary employment lawyer Roxanne Davis warned against "filtering people out for things in their resume that might reveal a gender, or family or race" in an April 2026 report on legal hazards facing Canadian recruiters. "Anything like that would run afoul of human rights legislation," she told Canadian HR Reporter.

"As AI increasingly informs professional decision-making, understanding how these systems reflect and shape human preferences and aspirations will be critical for ensuring that technological progress advances, rather than constrains, equal opportunity," the authors concluded.

Table 2: Workforce and regulatory context