PsychoLex: Unveiling the Psychological Mind of Large Language Models

Mohammad Amin Abbasi,Farnaz Sadat Mirnezami,Hassan Naderi

2024-08-17

Abstract:This paper explores the intersection of psychology and artificial intelligence through the development and evaluation of specialized Large Language Models (LLMs). We introduce PsychoLex, a suite of resources designed to enhance LLMs' proficiency in psychological tasks in both Persian and English. Key contributions include the PsychoLexQA dataset for instructional content and the PsychoLexEval dataset for rigorous evaluation of LLMs in complex psychological scenarios. Additionally, we present the PsychoLexLLaMA model, optimized specifically for psychological applications, demonstrating superior performance compared to general-purpose models. The findings underscore the potential of tailored LLMs for advancing psychological research and applications, while also highlighting areas for further refinement. This research offers a foundational step towards integrating LLMs into specialized psychological domains, with implications for future advancements in AI-driven psychological practice.

Computation and Language

What problem does this paper attempt to address?

### Problems the Paper Aims to Solve

This paper aims to address the application issues of large language models (LLMs) in the field of psychology. Specifically:

1. **Lack of datasets in the field of psychology**: Currently, there is a lack of high-quality datasets specifically targeted at the field of psychology, which limits the practical application of LLMs in psychological research, therapy, and education.

2. **Limitations of existing models in specialized fields**: Although general-purpose LLMs perform well in natural language processing, they fall short when dealing with complex psychological issues, particularly in terms of accuracy in understanding and responding to complex psychological queries.

To address these issues, the authors have developed the following resources:
- **PsychoLexQA dataset**: A dataset of instructional content and detailed Q&A in the field of psychology for training LLMs.
- **PsychoLexEval dataset**: A multiple-choice question dataset for evaluating the performance of LLMs in complex psychological scenarios.
- **PsychoLexLLaMA model**: An LLM specifically optimized for psychological tasks, with improved performance in psychological tasks through continued pre-training and fine-tuning.

Through these resources and methods, the paper demonstrates the potential applications of specially designed LLMs in the field of psychology and provides a foundation for further research.