BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//132.216.98.100//NONSGML kigkonsult.se iCalcreator 2.20.4//
BEGIN:VEVENT
UID:20260403T233416EDT-11995jLknb@132.216.98.100
DTSTAMP:20260404T033416Z
DESCRIPTION:Charles C.Y. Wang\n\nTandon Family Professor of Business Admini
 stration at Harvard Business Schoo\n\nWhen LLMs Go Abroad: Foreign Bias in
  AI Financial Predictions\n\nDate: Friday\, December 19\, 2025\n	Time: 10:3
 0 AM – 12:00 PM\n	Location: Bronfman building\, Room 245\n\n\nAbstract\n\nW
 e document foreign biases in AI-generated financial predictions: ChatGPT (
 US-based) is systematically more optimistic about Chinese firms than DeepS
 eek (China-based)\, predicting higher end-of-year stock prices and generat
 ing more buy recommendations. This AI-specific phenomenon contradicts the 
 traditional home bias in which investors favor domestic assets. We trace t
 his bias to differential information access: ChatGPT's optimism increases 
 when US media coverage of Chinese firms' negative news is scarce relative 
 to Chinese media. Supporting this mechanism\, placebo tests with synthetic
  Chinese firms without such asymmetries show no prediction gap between mod
 els. Crucially\, providing ChatGPT with Chinese news through prompts-which
  cannot alter model weights-completely eliminates the prediction gap\, dem
 onstrating that the bias stems from missing training data. Our findings im
 ply that the parallel development of LLMs in different countries can creat
 e divergent financial forecasts\, potentially amplifying rather than reduc
 ing cross-border information asymmetries as these tools shape investment d
 ecisions globally.\n
DTSTART:20251219T153000Z
DTEND:20251219T170000Z
LOCATION:Room 245\, Bronfman Building\, CA\, QC\, Montreal\, H3A 1G5\, 1001
  rue Sherbrooke Ouest
SUMMARY:Accounting Academic Area Workshop Series: Prof. Charles C.Y. Wang
URL:https://www.mcgill.ca/desautels/channels/event/accounting-academic-area
 -workshop-series-prof-charles-cy-wang-369809
END:VEVENT
END:VCALENDAR
