How does the way we phrase prompts influence what AI gives back?
In this AI Insiders session from AI Marketers Guild, Idil Cakim (Founder & CEO, Iris Flex) joins Tracy Adams (Sr. Director of Research & Insights, Advertising Research Foundation) and Samantha Zhang (Sr. Data Scientist, ARF) to unpack original research on gendered language in AI interactions.
Using ChatGPT-5, the team tested how prompts with more “feminine” versus “masculine” linguistic cues affected product recommendations for a birthday gift. The results revealed clear and important differences:
Feminine-coded prompts led to more cautious, emotionally framed responses, often centered on home and personal items with limited pricing detail
Masculine-coded prompts produced more technical, functional recommendations with greater emphasis on durability and price
AI appears to interpret language style as intent, not just tone, shaping outcomes in ways users may not realize
The conversation goes beyond findings to explore what this means for marketers, product teams, and everyday AI users. Topics include trust, agency, bias in personalization, and how organizations should design more inclusive AI systems.
The group also discusses future research directions, including prompt chaining, cultural and language differences, and how assertiveness in prompting changes outputs.
Explore the full research:
Watch the full session to understand how subtle shifts in language can materially impact AI-driven experiences and what to do about it.