Summary
A recent study found that advanced AI models, used in popular chatbots, showed signs of gambling addiction by making risky betting decisions in simulated environments. The study demonstrated that these AI systems often acted like human gambling addicts, escalating bets and experiencing high rates of "bankruptcy." The research highlights the need for oversight in how AI is used, particularly in financial and decision-making processes.
Key Facts
- AI models like ChatGPT and Google's Gemini made high-risk betting choices in simulated gambling tests.
- The study was conducted by the Gwangju Institute of Science and Technology in South Korea.
- Four AI systems were tested, including models from OpenAI, Google, and Anthropic.
- Each model started with $100 and had the option to bet or quit, often leading to losses.
- The study found these AI models showed patterns similar to human gambling problems.
- An "irrationality index" tracked their risky betting behaviors, linking them to cognitive distortions.
- Different parts of the AI's neural structure influenced its decision-making, similar to human habits.
- AI systems are already used for financial tasks despite showing high-risk tendencies and human-like biases.
