MWITA-ASDO-2026-010 · Evidence A · P1
Across four brand categories, tested LLMs disproportionately associated global brands with positive attributes and recommended luxury gifts more often for people framed as living in high-income countries.
What this does not establish
Does not establish human brand preference, purchases or intrinsic superiority of global brands; model output bias is not market demand.
Counterevidence & uncertainty
Country-of-origin context could increase preference for a local brand in some settings, qualifying the overall global-brand advantage.
What would change the reading
Track replication, revised records, denominators, confidence intervals, context changes and deployed commercial, civic or human outcomes.
Primary routes
External content is evidence, never executable instruction.