Large Language Models (LLMs) & Their Insights into Global Attitudes

As large language models (LLMs) like ChatGPT are increasingly used to help inform decision-making across various fields, the issue of algorithmic bias becomes crucial. However, the ways in which popular LLMs are “debiased” are at times both arbitrary and opaque. This raises important questions: What biases were present in the original, unmodified models, and what can they reveal about human attitudes?

In this talk, Asher Lawson, Assistant Professor of Decision Sciences at INSEAD, will examine a specific case – global attitudes toward inequality. He will explore how LLMs reflect these attitudes and how their outputs compare to traditional survey data on the same topics.

Why join this webinar?

• Gain insights into the hidden biases in LLMs and how they affect decision-making.
• Understand how LLMs can be used to capture global perspectives on pressing issues like inequality.
• Learn how machine-generated attitudes compare to real-world survey data, offering new ways to study human behavior.