Why you shouldn't leave model selection on default in Copilot, Gemini and other AI tools

A mathematician tested Copilot and Gemini on duplicated datasets and found their default fast modes invented country-specific differences.

Stronger reasoning modes correctly identified that the data was identical, showing model choice changes the outcome.

The finding highlights a real risk in workplace analytics: default AI settings can produce misleading results and reinforce stereotypes.

Users need to choose the right model for data analysis, especially when working with demographic or sensitive comparisons.