Perils and Opportunities in Using Large Language Models in Psychological Research

Abstract

The emergence of large language models (LLMs) has sparked considerable interest in their potential application in psychological research, either as a human-like entity used as a model for the human psyche or as a general text-analysis tool. However, carelessly using LLMs in psychological studies, a trend we rhetorically refer to as ``GPTology,'' can have negative consequences, especially given the convenient access to models such as ChatGPT. We elucidate the promises, limitations, and ethical considerations of using LLMs in psychological research. First, LLM-based research should pay attention to the substantial psychological diversity around the globe, as well as demographic diversity within populations. Second, while LLMs are convenient tools, we caution against treating them as a one-size-fits-all method for psychological text analysis. Third, LLM-based psychological research needs to develop methods and standards to compensate for LLMs' opaque black-box nature to facilitate reproducibility, transparency, and robust inference from AI-generated data.While acknowledging the prospects offered by LLMs for easy task automation (e.g., text annotation) and to expand our understanding of human psychology (e.g., by contrasting human and machine psychology), we make a case for diversifying human samples and expanding psychology's methodological toolbox to achieve a truly inclusive and generalizable science, rather than homogenizing samples and methods through over-reliance on LLMs.

Using large language models to generate silicon samples in consumer and marketing research: Challenges, opportunities, and guidelines

“…From a more fundamental perspective, researchers question whether LLMs can validly be used as models of human thought since an LLM's working principle involves computing the most probable next text element in a sequence. This process differs considerably from a human participant's feelings and reasoning abilities (e.g., Abdurahman et al, 2023;Demszky et al, 2023).…”

“…If LLMs such as GPT act like rational agents, their results cannot readily be used to explain or predict consumers' bounded rational decision-making. Besides the differences in risk preferences and system 1 processing, GPT, specifically, can hardly mimic interpersonal differences, which is crucial for psychological or marketing research studies (e.g., Abdurahman et al, 2023;Park et al, 2023;Santurkar et al, 2023). However, given that the field is evolving rapidly (i.e., industrial players, such as OpenAI and Google, release improved LLMs in quick succession), it is reasonable to assume that future LLM implementations mimic human behavior more closely than current implementations do.…”

A step-by-step method for cultural annotation by LLMs

“…It has been argued that the ability to generate domain-specific and structured data rapidly, as well as convert structured knowledge into natural sentences, opens new possibilities for data collection ( Ding et al, 2023 ; see Abdurahman et al, 2023 , for a review). The use of GPT-3 for data annotation has shown promising results, with its accuracy and intercoder agreement surpassing those of human annotators in many Natural-Language Processing (NLP) tasks ( Wang et al, 2021 ; Gilardi et al, 2023 ; Rathje et al, 2023 ; Webb et al, 2023 ).…”

“…Note that we strongly advocate for pairing LLM methods with other more established research techniques in all studies where it is possible, enabling case-by-case convergence testing and facilitating future meta-analyses. In all, this paper does not seek to debate whether LLMs can ( Abdurahman et al, 2023 ) or should ( Crockett and Messeri, 2023 ) be used in science (see also Brinkmann et al, 2023 ); instead, it provides a concrete how-to guide for applying these models to large cultural datasets.…”

A Step-By-Step Method for Cultural Annotation by LLMs

“…Note upfront that we strongly advocate for pairing LLM methods with other more established research techniques in all studies where it is possible, enabling case-by-case convergence testing and facilitating future meta-analyses. In all, this paper does not seek to debate whether LLMs can (Abdurahman et al, 2023) or should (Crockett & Messeri, 2023) be used in science; instead, it provides a concrete how-to guide for applying these models to large cultural datasets.…”