Inducing anxiety in large language models increases exploration and bias. (arXiv:2304.11111v1 [cs.CL])

Large language models are transforming research on machine learning while
galvanizing public debates. Understanding not only when these models work well
and succeed but also why they fail and misbehave is of great societal
relevance. We propose to turn the lens of computational psychiatry, a framework
used to computationally describe and modify aberrant behavior, to the outputs
produced by these models. We focus on the Generative Pre-Trained Transformer
3.5 and subject it to tasks commonly studied in psychiatry. Our results …

anxiety arxiv behavior bias computational exploration focus framework generative generative pre-trained transformer gpt gpt-3 gpt-3.5 language language models large language models machine machine learning public research transformer understanding work