Despite ongoing efforts to close the gender wage gap in Canada, women still earn approximately 17 percent less than men. Globally, progress is even slower, with the World Economic Forum estimating it will take 169 years to achieve parity in economic opportunity between genders. As women increasingly turn to artificial intelligence (AI) tools to navigate salary negotiations and compensation decisions, concerns are emerging that these technologies may inadvertently perpetuate or even exacerbate existing disparities.

AI language models such as ChatGPT, Claude, and Gemini have become widely used for a variety of tasks, including career-related inquiries. OpenAI reports that Americans send nearly three million daily messages to ChatGPT concerning wages and compensation, underscoring the reliance on AI for salary guidance. However, experts warn these models reflect the biases embedded in their training data, which often mirrors systemic inequalities.

Danielle Gifford, managing director of AI and advanced analytics at PricewaterhouseCoopers in Calgary, explains that large language models (LLMs) undergo an initial training phase using vast amounts of internet text, followed by human-guided fine-tuning to improve helpfulness and safety. Because these models learn from historical data where gender biases persist, they may reproduce discriminatory patterns. A 2025 study by the Association for Computational Linguistics found that LLMs recommended lower salary targets for female personas compared to identical male counterparts, confirming the presence of gender bias in AI outputs.

Negotiations expert Fotini Iconomopoulos, based in Toronto, suggests that AI should be used cautiously as a research tool rather than an authoritative adviser. She encourages users to apply critical judgment and independently verify AI-generated information rather than accepting it at face value. Ms. Gifford adds that AI responses vary depending on perceived user gender unless explicitly corrected, highlighting the importance of carefully framing requests to avoid reinforcing stereotypes.

Several users have found value in AI assistance despite its limitations. Abigail Shakespeare, an executive in British Columbia’s charitable sector, leveraged ChatGPT to restructure her resume and identify salary benchmarks in her industry. Though aware of AI’s biases, she combined the tool’s insights with independent research on salary data from sources such as Glassdoor and Canada Revenue Agency filings. This approach enabled her to negotiate a significant pay increase—from an initial offer of $90,000 to a final compensation package totaling $150,000, including bonuses.

Still, experts caution that not all users take these extra steps. Because AI suggestions may anchor salary expectations lower for women, there is a risk that some individuals will accept suboptimal offers, compounding lifetime income and wealth disparities. A hypothetical $60,000 compensation gap maintained over 20 years can translate into more than $2.5 million in lost investment growth and diminished financial security.

Gifford notes the absence of regulatory frameworks in Canada to address AI bias and calls on technology companies to improve model design by disentangling attributes such as gender, age, and race from training data. Meanwhile, women remain underrepresented in AI development, comprising less than 30 percent of personnel, which limits efforts to identify and mitigate bias from within.

While AI alone will not solve the gender wage gap, industry experts emphasize that women benefit from engaging with these tools proactively and critically. Masking personal identifiers when consulting AI, cross-referencing outputs with external data, and advocating based on informed insights can help counterbalance entrenched biases embedded in the technology. The challenge remains to shape AI as a resource that supports equitable outcomes rather than unwittingly reinforcing existing inequalities.