Generative Artificial Intelligence. Potential Effects on Psychological Research

Abstract

Spurred by the launch of transformer models like GPT, generative AI (Artificial Intelligence) has made enormous strides. This position paper identifies two core features that turn generative AI into an exciting topic and instrument for psychological research: the scope of use cases and the ability to create human-like output. Against this background, the paper discusses potential implications for psychological research in three areas. First, we observe a burgeoning trend to analyze the abilities of generative AI from a psychological perspective. The overarching question in this field is whether the output and behavior of generative AI exhibit human-like properties. Second, psychological research will play an essential role in assessing the impact of generative AI on humans. Key questions are about how generative AI can be used deceptively, how humans evaluate AI regarding competence or warmth, and how generative AI can be leveraged to foster knowledge-related processes in interaction with humans. Third, the process of psychological research itself is likely to change with generative AI. We discuss potential fields of application but also argue against some use cases (e. g., employing generative AI as a surrogate for participants in experiments). In all three research fields, the dualism between the lack of language understanding, coupled with an uncanny ability to (re) produce human-like language, is both a challenge for and a motivator of psychological research.