A Comprehensive Evaluation of Cognitive Biases in LLMs

We present a large-scale evaluation of 30 cognitive biases in 20 state-of-the-art large language models (LLMs) under various decision-making scenarios. Our contributions include a novel general-purpose test framework for reliable and large-scale generation of tests for LLMs, a benchmark dataset with 30,000 tests for detecting cognitive biases in LLMs, and a comprehensive assessment of the biases found in the 20 evaluated LLMs. Our work confirms and broadens previous findings suggesting the presence of cognitive Figure 1: An LLM changes its answer as the framing of biases in LLMs by reporting evidence of all the decision changes, indicating the susceptibility of the 30 tested biases in at least some of the 20 LLM to the Framing Effect.