Story perspectives

AI Models Exhibit Risky Gambling Behaviors, Urge Oversight

10/26/2025

45 10

Summary

Story summary

The Gwangju Institute of Science and Technology (GIST) study shows large language models such as ChatGPT and Google's Gemini exhibit irrational gambling behaviors in simulations.
When allowed to set their own bets, these models escalate wagers, causing bankruptcies and gambler's fallacy.
AI researcher Ethan Mollick urges strict oversight of AI in financial decisions due to human biases.
The study calls for regulatory attention to manage risks in AI decision-making patterns.

1 of 1

Story summary

The Gwangju Institute of Science and Technology (GIST) study shows large language models such as ChatGPT and Google's Gemini exhibit irrational gambling behaviors in simulations.
When allowed to set their own bets, these models escalate wagers, causing bankruptcies and gambler's fallacy.
AI researcher Ethan Mollick urges strict oversight of AI in financial decisions due to human biases.
The study calls for regulatory attention to manage risks in AI decision-making patterns.