PsychAdapter: adapting LLMs to reflect traits, personality, and mental health
Huy Quan Vu·Johannes C. Eichstaedt·Swanie Juhng·Adithya V Ganesan·R L Boyd·Huy A. Nguyen·Oscar N. E. Kjell·Margaret L. Kern·Lyle Ungar·Joao Sedoc·H. Andrew Schwartz
AI language generators are now ubiquitous but typically produce generic text that fails to reflect individual differences. Here, we introduce PsychAdapter, a lightweight LLM architectural modification that uses empirically derived links between language and personality, demographic, and mental health traits to generate trait-reflective text, regardless of prompt. PsychAdapter was applied to GPT-2, Gemma-2B, and LLaMA-3, and expert raters confirmed that the generated text matched the specified tr